This paper addresses the problem of reconstructing a scene online at the level of objects given an RGB-D video sequence. While current object-aware neural implicit rep-resentations hold promise, they are limited in online reconstruction efficiency and shape completion. Our main contributions to alleviate the above limitations are twofold. First, we propose a feature grid interpolation mechanism to continuously update grid-based object-centric neural implicit representations as new object parts are revealed. Second, we construct an object library with previously mapped objects in advance and leverage the corresponding shape priors to initialize geometric object models in new videos, sub-sequently completing them with novel views as well as synthesized past views to avoid losing original object details. Extensive experiments on synthetic environments from the Replica dataset, real-world ScanNet sequences and videos captured in our laboratory demonstrate that our approach outperforms state-of-the-art neural implicit models for this task in terms of reconstruction accuracy and completeness.

The visual information-seeking task aims to answer visual questions that require external knowledge, such as “On what date did this building officially open?”. Existing methods using retrieval-augmented generation framework primarily rely on textual knowledge bases to assist multimodal large language models (MLLMs) in answering questions. However, the text-only knowledge can impair information retrieval for the multimodal query of image and question, and also confuse MLLMs in selecting the most relevant information during generation. In this work, we propose a novel framework MuKA which leverages a multimodal knowledge base to address these limitations. Specifically, we construct a multimodal knowledge base by automatically pairing images with text passages in existing datasets. We then design a fine-grained multimodal interaction to effectively retrieve multimodal documents and enrich MLLMs with both retrieved texts and images. MuKA outperforms state-of-the-art methods by 38.7% and 15.9% on the InfoSeek and E-VQA benchmark respectively, demonstrating the importance of multimodal knowledge in enhancing both retrieval and answer generation.

Reconstructing hand-held objects in 3D from monocular images remains a significant challenge in computer vision. Most existing approaches rely on implicit 3D representations, which produce overly smooth reconstructions and are time-consuming to generate explicit 3D shapes. While more recent methods directly reconstruct point clouds with diffusion models, the multi-step denoising makes high-resolution reconstruction inefficient. To address these limitations, we propose a transformer-based model to efficiently reconstruct dense 3D point clouds of hand-held objects. Our method follows a coarse-to-fine strategy, first generating a sparse point cloud from the image and progressively refining it into a dense representation using pixel-aligned image features. To enhance reconstruction accuracy, we integrate image features with 3D hand geometry to jointly predict the object point cloud and its pose relative to the hand. Our model is trained end-to-end for optimal performance. Experimental results on both synthetic and real datasets demonstrate that our method achieves state-of-the-art accuracy with much faster inference speed, while generalizing well to in-the-wild images.

This work addresses the problem of active 3D mapping, where an agent must find an efficient trajectory to exhaustively reconstruct a new scene. Previous approaches mainly predict the next best view near the agent's location, which is prone to getting stuck in local areas. Additionally, existing indoor datasets are insufficient due to limited geometric complexity and inaccurate ground truth meshes. To overcome these limitations, we introduce a novel dataset AiMDoom with a map generator for the Doom video game, enabling to better benchmark active 3D mapping in diverse indoor environments. Moreover, we propose a new method we call next-best-path (NBP), which predicts long-term goals rather than focusing solely on short-sighted views. The model jointly predicts accumulated surface coverage gains for long-term goals and obstacle maps, allowing it to efficiently plan optimal paths with a unified model. By leveraging online data collection, data augmentation and curriculum learning, NBP significantly outperforms state-of-the-art methods on both the existing MP3D dataset and our AiMDoom dataset, achieving more efficient mapping in indoor environments of varying complexity.

In this work, we aim to learn a unified vision-based policy for multi-fingered robot hands to manipulate a variety of objects in diverse poses. Though prior work has shown benefits of using human videos for policy learning, performance gains have been limited by the noise in estimated trajectories. Moreover, reliance on privileged object information such as ground-truth object states further limits the applicability in realistic scenarios. To address these limitations, we propose a new framework ViViDex to improve vision-based policy learning from human videos. It first uses reinforcement learning with trajectory guided rewards to train state-based policies for each video, obtaining both visually natural and physically plausible trajectories from the video. We then rollout successful episodes from state-based policies and train a unified visual policy without using any privileged information. We propose coordinate transformation to further enhance the visual point cloud representation, and compare behavior cloning and diffusion policy for the visual policy training. Experiments both in simulation and on the real robot demonstrate that ViViDex outperforms state-of-theart approaches on three dexterous manipulation tasks. Project website: zerchen.github.io/projects/vividex.html.

Generalizing language-conditioned robotic policies to new tasks remains a significant challenge, hampered by the lack of suitable simulation benchmarks. In this paper, we address this gap by introducing GemBench, a novel benchmark to assess generalization capabilities of vision-language robotic manipulation policies. GemBench incorporates seven general action primitives and four levels of generalization, spanning novel placements, rigid and articulated objects, and complex long-horizon tasks. We evaluate state-of-the-art approaches on GemBench and also introduce a new method. Our approach 3D- LOTUS leverages rich 3D information for action prediction conditioned on language. While 3D-LOTUS excels in both efficiency and performance on seen tasks, it struggles with novel tasks. To address this, we present 3D-LOTUS++, a framework that integrates 3D-LOTUS's motion planning capabilities with the task planning capabilities of LLMs and the object grounding accuracy of VLMs. 3D-LOTUS++ achieves state-of-the-art per-formance on novel tasks of GemBench, setting a new standard for generalization in robotic manipulation. Code, dataset, real robot videos and trained models are available at https://www.di.ens.fr/willow/research/gembench/.

Learning generalizable visual representations from Internet data has yielded promising results for robotics. Yet, prevailing approaches focus on pre-training 2D representations, being sub-optimal to deal with occlusions and accurately localize objects in complex 3D scenes. Meanwhile, 3D representation learning has been limited to single-object understanding. To address these limitations, we introduce a novel 3D pre-training framework for robotics named SUGAR that captures semantic, geometric and affordance properties of objects through 3D point clouds. We underscore the importance of cluttered scenes in 3D representation learning, and automatically construct a multi-object dataset benefiting from cost-free supervision in simulation. SUGAR employs a versatile transformer-based model to jointly address five pre-training tasks, namely cross-modal knowledge distillation for semantic learning, masked point modeling to understand geometry structures, grasping pose synthesis for object affordance, 3D instance segmentation and referring expression grounding to analyze cluttered scenes. We evaluate our learned representation on three robotic-related tasks, namely, zero-shot 3D object recognition, referring expression grounding, and language-driven robotic manipulation. Experimental results show that SUGAR's 3D representation outperforms state-of-the-art 2D and 3D representations.

Abstract Salient Object Detection (SOD) is a challenging task that aims to precisely identify and segment the salient objects. However, existing SOD methods still face challenges in making explicit predictions near the edges and often lack end‐to‐end training capabilities. To alleviate these problems, we propose SOD‐diffusion, a novel framework that formulates salient object detection as a denoising diffusion process from noisy masks to object masks. Specifically, object masks diffuse from ground‐truth masks to random distribution in latent space, and the model learns to reverse this noising process to reconstruct object masks. To enhance the denoising learning process, we design an attention feature interaction module (AFIM) and a specific fine‐tuning protocol to integrate conditional semantic features from the input image with diffusion noise embedding. Extensive experiments on five widely used SOD benchmark datasets demonstrate that our proposed SOD‐diffusion achieves favorable performance compared to previous well‐established methods. Furthermore, leveraging the outstanding generalization capability of SOD‐diffusion, we applied it to publicly available images, generating high‐quality masks that serve as an additional SOD benchmark testset.

Automatic image captioning evaluation is critical for benchmarking and promoting advances in image captioning research.Existing metrics only provide a single score to measure caption qualities, which are less explainable and informative.Instead, we humans can easily identify the problems of captions in details, e.g., which words are inaccurate and which salient objects are not described, and then rate the caption quality.To support such informative feedback, we propose an Informative Metric for Reference-free Image Caption evaluation (InfoMetIC).Given an image and a caption, InfoMetIC is able to report incorrect words and unmentioned image regions at fine-grained level, and also provide a text precision score, a vision recall score and an overall quality score at coarse-grained level.The coarse-grained score of InfoMetIC achieves significantly better correlation with human judgements than existing metrics on multiple benchmarks.We also construct a token-level evaluation dataset and demonstrate the effectiveness of InfoMetIC in fine-grained evaluation.Our code and datasets are publicly available at https://github. com/HAWLYQ/InfoMetIC.

Signed distance functions (SDFs) is an attractive frame-work that has recently shown promising results for 3D shape reconstruction from images. SDFs seamlessly generalize to different shape resolutions and topologies but lack explicit modelling of the underlying 3D geometry. In this work, we ex-ploit the hand structure and use it as guidance for SDF-based shape reconstruction. In particular, we address reconstruction of hands and manipulated objects from monocular RGB images. To this end, we estimate poses of hands and objects and use them to guide 3D reconstruction. More specifically, we predict kinematic chains of pose transformations and align SDFs with highly-articulated hand poses. We improve the visual features of 3D points with geometry alignment and further leverage temporal information to enhance the robustness to occlusion and motion blurs. We conduct extensive experiments on the challenging ObMan and DexYCB benchmarks and demonstrate significant improvements of the proposed method over the state of the art.

While current visual captioning models have achieved impressive performance, they often assume that the image is well-captured and provides a complete view of the scene. In real-world scenarios, however, a single image may not offer a good viewpoint, hindering fine-grained scene understanding. To overcome this limitation, we propose a novel task called Embodied Captioning, which equips visual captioning models with navigation capabilities, enabling them to actively explore the scene and reduce visual ambiguity from suboptimal viewpoints. Specifically, starting at a random viewpoint, an agent must navigate the environment to gather information from different viewpoints and generate a comprehensive paragraph describing all objects in the scene. To support this task, we build the ET-Cap dataset with Kubric simulator, consisting of 10K 3D scenes with cluttered objects and three annotated paragraphs per scene. We propose a Cascade Embodied Captioning model (CaBOT), which comprises of a navigator and a captioner, to tackle this task. The navigator predicts which actions to take in the environment, while the captioner generates a paragraph description based on the whole navigation trajectory. Extensive experiments demonstrate that our model outperforms other carefully designed baselines. Our dataset, codes and models are available at https://aim3-ruc.github.io/ExploreAndTell.

Object goal navigation aims to navigate an agent to locations of a given object category in unseen environments. Classical methods explicitly build maps of environments and require extensive engineering while lacking semantic information for object-oriented exploration. On the other hand, end-to-end learning methods alleviate manual map design and predict actions using implicit representations. Such methods, however, lack an explicit notion of geometry and may have limited ability to encode navigation history. In this work, we propose an implicit spatial map for object goal navigation. Our implicit map is recursively updated with new observations at each step using a transformer. To encourage spatial reasoning, we introduce auxiliary tasks and train our model to reconstruct explicit maps as well as to predict visual features, semantic labels and actions. Our method significantly outperforms the state of the art on the challenging MP3D dataset and generalizes well to the HM3D dataset. We successfully deploy our model on a real robot and achieve encouraging object goal navigation results in real scenes using only a few real-world demonstrations. Code, trained models and videos are available at https://www.di.ens.fr/willow/research/onav_rim/.

Learning visuomotor policies in simulation is much safer and cheaper than in the real world. However, due to discrepancies between the simulated and real data, simulator-trained policies often fail when transferred to real robots. One common approach to bridge the visual sim-to-real domain gap is domain randomization (DR). While previous work mainly evaluates DR for disembodied tasks, such as pose estimation and object detection, here we systematically explore visual domain randomization methods and benchmark them on a rich set of challenging robotic manipulation tasks. In particular, we propose an offline proxy task of cube localization to select DR parameters for texture randomization, lighting randomization, variations of object colors and camera parameters. Notably, we demonstrate that DR parameters have similar impact on our offline proxy task and online policies. We, hence, use offline optimized DR parameters to train visuomotor policies in simulation and directly apply such policies to a real robot. Our approach achieves 93% success rate on average when tested on a diverse set of challenging manipulation tasks. Moreover, we evaluate the robustness of policies to visual variations in real scenes and show that our simulator-trained policies outperform policies learned using real but limited data. Code, simulation environment, real robot datasets and trained models are available at https://www.di.ens.fr/willow/research/robust_s2r/.

A video storyboard is a roadmap for video creation which consists of shot-by-shot images to visualize key plots in a text synopsis. Creating video storyboards, however, remains challenging which not only requires cross-modal association between high-level texts and images but also demands long-term reasoning to make transitions smooth across shots. In this paper, we propose a new task called Text synopsis to Video Storyboard (TeViS) which aims to retrieve an ordered sequence of images as the video storyboard to visualize the text synopsis. We construct a MovieNet-TeViS dataset based on the public MovieNet dataset [17]. It contains 10K text synopses each paired with keyframes manually selected from corresponding movies by considering both relevance and cinematic coherence. To benchmark the task, we present strong CLIP-based baselines and a novel VQ-Trans model. VQ-Trans first encodes text synopsis and images into a joint embedding space and uses vector quantization (VQ) to improve the visual representation. Then, it auto-regressively generates a sequence of visual features for retrieval and ordering. Experimental results demonstrate that VQ-Trans significantly outperforms prior methods and the CLIP-based baselines. Nevertheless, there is still a large gap compared to human performance suggesting room for promising future work. The code and data are available at: https://ruc-aimind.github.io/projects/TeViS/

Following language instructions to navigate in unseenaenvironments is a challenging problem for autonomous embodied agents. The agent not only needs to ground languages in visual scenes, but also should explore the environment to reach its target. In this work, we propose a dual-scale graph transformer (DUET) for joint long-term action planning and fine-grained cross-modal understanding. We build a topological map on-the-fly to enable efficient exploration in global action space. To balance the complexity of large action space reasoning and fine-grained language grounding, we dynamically combine a fine-scale encoding over local observations and a coarse-scale encoding on a global map via graph transformers. The proposed approach, DUET, significantly outperforms state-of-the-art methods on goal-oriented vision-and-language navigation (VLN) benchmarks REVERIE and SOON. It also improves the success rate on the fine-grained VLN benchmark R2R.

Visual relation understanding plays an essential role for holistic video understanding. Most previous works adopt a multi-stage framework for video visual relation detection (VidVRD), which cannot capture long-term spatio-temporal contexts in different stages and also suffers from inefficiency. In this paper, we propose a transformer-based framework called VRDFormer to unify these decoupling stages. Our model exploits a query-based approach to autoregressively generate relation instances. We specifically design static queries and recurrent queries to enable efficient object pair tracking with spatio-temporal contexts. The model is jointly trained with object pair detection and relation classification. Extensive experiments on two benchmark datasets, ImageNet-VidVRD and VidOR, demonstrate the effectiveness of the proposed VRDFormer, which achieves the state-of-the-art performance on both relation detection and relation tagging tasks. The code is released at https://github.com/zhengsipeng/VRDFormer_VRD.

Localizing objects in 3D scenes based on natural language requires understanding and reasoning about spatial relations. In particular, it is often crucial to distinguish similar objects referred by the text, such as "the left most chair" and "a chair next to the window". In this work we propose a language-conditioned transformer model for grounding 3D objects and their spatial relations. To this end, we design a spatial self-attention layer that accounts for relative distances and orientations between objects in input 3D point clouds. Training such a layer with visual and language inputs enables to disambiguate spatial relations and to localize objects referred by the text. To facilitate the cross-modal learning of relations, we further propose a teacher-student approach where the teacher model is first trained using ground-truth object labels, and then helps to train a student model using point cloud inputs. We perform ablation studies showing advantages of our approach. We also demonstrate our model to significantly outperform the state of the art on the challenging Nr3D, Sr3D and ScanRefer 3D object grounding datasets.

Multimodal machine translation (MMT) aims to translate a sentence in the source language into the target language with the context of an associated image. According to where the visual information is employed, previous approaches can be categorized into two types: directly injecting the visual information at the input side or exploiting it as a visual constraint at the objective side. In this work, we propose an IO-MMT model which exploits the visual assistance in dual sides to fully exploit the visual information for multimodal machine translation. It contains a relation-aware multimodal transformer to simultaneously exploit the objects and their spatial relationships in the image with a graph at the input-side and a novel visual assistance structure to further improve visual consistency of the translation at the objective-side. Experimental results under both normal setting and input degradation settings on the Multi30k benchmark dataset show that combining the visual assistance in dual sides consistently outperforms single-side MMT models and achieves the state-of-the-art results on EN-DE and EN-FR translation tasks. We will release the codes and models at https://github.com/syuqings/MMT.

Video paragraph captioning aims to describe multiple events in untrimmed videos with descriptive paragraphs. Existing approaches mainly solve the problem in two steps: event detection and then event captioning. Such two-step manner makes the quality of generated paragraphs highly dependent on the accuracy of event proposal detection which is already a challenging task. In this paper, we propose a paragraph captioning model which eschews the problematic event detection stage and directly generates paragraphs for untrimmed videos. To describe coherent and diverse events, we propose to enhance the conventional temporal attention with dynamic video memories, which progressively exposes new video features and suppresses over-accessed video contents to control visual focuses of the model. In addition, a diversity-driven training strategy is proposed to improve diversity of paragraph on the language perspective. Considering that untrimmed videos generally contain massive but redundant frames, we further augment the video encoder with keyframe awareness to improve efficiency. Experimental results on the ActivityNet and Charades datasets show that our proposed model significantly outperforms the state-of-the-art performance on both accuracy and diversity metrics without using any event boundary annotations. Code will be released at https://github.com/syuqings/video-paragraph.

The dense video captioning task aims to detect and describe a sequence of events in a video for detailed and coherent storytelling. Previous works mainly adopt a "detect-then-describe" framework, which firstly detects event proposals in the video and then generates descriptions for the detected events. However, the definitions of events are diverse which could be as simple as a single action or as complex as a set of events, depending on different semantic con-texts. Therefore, directly detecting events based on video information is ill-defined and hurts the coherency and accuracy of generated dense captions. In this work, we reverse the predominant "detect-then-describe" fashion, proposing a top-down way to first generate paragraphs from a global view and then ground each event description to a video segment for detailed refinement. It is formulated as a Sketch, Ground, and Refine process (SGR). The sketch stage first generates a coarse-grained multi-sentence paragraph to describe the whole video, where each sentence is treated as an event and gets localised in the grounding stage. In the re-fining stage, we improve captioning quality via refinement-enhanced training and dual-path cross attention on both coarse-grained event captions and aligned event segments. The updated event caption can further adjust its segment boundaries. Our SGR model outperforms state-of-the-art methods on ActivityNet Captioning benchmark under traditional and story-oriented dense caption evaluations. Code will be released at github.com/bearcatt/SGR.

The growing number of action classes has posed a new challenge for video understanding, making Zero-Shot Action Recognition (ZSAR) a thriving direction. The ZSAR task aims to recognize target (unseen) actions without training examples by leveraging semantic representations to bridge seen and unseen actions. However, due to the complexity and diversity of actions, it remains challenging to semantically represent action classes and transfer knowledge from seen data. In this work, we propose an ER-enhanced ZSAR model inspired by an effective human memory technique Elaborative Rehearsal (ER), which involves elaborating a new concept and relating it to known concepts. Specifically, we expand each action class as an Elaborative Description (ED) sentence, which is more discriminative than a class name and less costly than manual-defined attributes. Besides directly aligning class semantics with videos, we incorporate objects from the video as Elaborative Concepts (EC) to improve video semantics and generalization from seen actions to unseen actions. Our ER-enhanced ZSAR model achieves state-of-the-art results on three existing benchmarks. Moreover, we propose a new ZSAR evaluation protocol on the Kinetics dataset to overcome limitations of current benchmarks and first compare with few-shot learning baselines on this more realistic setting. Our codes and collected EDs are released at https://github.com/DeLightCMU/ElaborativeRehearsal.

Vision-and-language navigation (VLN) aims to enable embodied agents to navigate in realistic environments using natural language instructions. Given the scarcity of domain-specific training data and the high diversity of image and language inputs, the generalization of VLN agents to unseen environments remains challenging. Recent methods explore pretraining to improve generalization, however, the use of generic image-caption datasets or existing small-scale VLN environments is suboptimal and results in limited improvements. In this work, we introduce BnB1, a large-scale and diverse in-domain VLN dataset. We first collect image-caption (IC) pairs from hundreds of thousands of listings from online rental marketplaces. Using IC pairs we next propose automatic strategies to generate millions of VLN path-instruction (PI) pairs. We further propose a shuffling loss that improves the learning of temporal order inside PI pairs. We use BnB to pretrain our Airbert2model that can be adapted to discriminative and generative settings and show that it outperforms state of the art for Room-to-Room (R2R) navigation and Remote Referring Expression (REVERIE) benchmarks. Moreover, our in-domain pretraining significantly increases performance on a challenging few-shot VLN evaluation, where we train the model only on VLN instructions from a few houses.

Pre-training has been an emerging topic that provides a way to learn strong representation in many fields (e.g., natural language processing, computing vision). In the last few years, we have witnessed many research works on multi-modal pre-training which have achieved state-of-the-art performances on many multimedia tasks (e.g., image-text retrieval, video localization, speech recognition). In this workshop, we aim to gather peer researchers on related topics for more insightful discussion. We also intend to attract more researchers to explore and investigate more opportunities of designing and using innovative pre-training models for multimedia tasks.

Translating e-commercial product descriptions, a.k.a product-oriented machine translation (PMT), is essential to serve e-shoppers all over the world. However, due to the domain specialty, the PMT task is more challenging than traditional machine translation problems. Firstly, there are many specialized jargons in the product description, which are ambiguous to translate without the product image. Secondly, product descriptions are related to the image in more complicated ways than standard image descriptions, involving various visual aspects such as objects, shapes, colors or even subjective styles. Moreover, existing PMT datasets are small in scale to support the research. In this paper, we first construct a large-scale bilingual product description dataset called Fashion-MMT, which contains over 114k noisy and 40k manually cleaned description translations with multiple product images. To effectively learn semantic alignments among product images and bilingual texts in translation, we design a unified product-oriented cross-modal cross-lingual model for pre-training and fine-tuning. Experiments on the Fashion-MMT and Multi30k datasets show that our model significantly outperforms the state-of-the-art models even pre-trained on the same dataset. It is also shown to benefit more from large-scale noisy data to improve the translation quality. We will release the dataset and codes at https://github.com/syuqings/Fashion-MMT.

For an image with multiple scene texts, different people may be interested in different text information. Current text-aware image captioning models are not able to generate distinctive captions according to various information needs. To explore how to generate personalized text-aware captions, we define a new challenging task, namely Question-controlled Text-aware Image Captioning (Qc-TextCap). With questions as control signals, this task requires models to understand questions, find related scene texts and describe them together with objects fluently in human language. Based on two existing text-aware captioning datasets, we automatically construct two datasets, ControlTextCaps and ControlVizWiz to support the task. We propose a novel Geometry and Question Aware Model (GQAM). GQAM first applies a Geometry-informed Visual Encoder to fuse region-level object features and region-level scene text features with considering spatial relationships. Then, we design a Question-guided Encoder to select the most relevant visual features for each question. Finally, GQAM generates a personalized text-aware caption with a Multimodal Decoder. Our model achieves better captioning performance and question answering ability than carefully designed baselines on both two datasets. With questions as control signals, our model generates more informative and diverse captions than the state-of-the-art text-aware captioning model. Our code and datasets are publicly available at https://github.com/HAWLYQ/Qc-TextCap.

Vision-and-language navigation (VLN) aims to build autonomous visual agents that follow instructions and navigate in real scenes. To remember previously visited locations and actions taken, most approaches to VLN implement memory using recurrent states. Instead, we introduce a History Aware Multimodal Transformer (HAMT) to incorporate a long-horizon history into multimodal decision making. HAMT efficiently encodes all the past panoramic observations via a hierarchical vision transformer (ViT), which first encodes individual images with ViT, then models spatial relation between images in a panoramic observation and finally takes into account temporal relation between panoramas in the history. It, then, jointly combines text, history and current observation to predict the next action. We first train HAMT end-to-end using several proxy tasks including single step action prediction and spatial relation prediction, and then use reinforcement learning to further improve the navigation policy. HAMT achieves new state of the art on a broad range of VLN tasks, including VLN with fine-grained instructions (R2R, RxR), high-level instructions (R2R-Last, REVERIE), dialogs (CVDN) as well as long-horizon VLN (R4R, R2R-Back). We demonstrate HAMT to be particularly effective for navigation tasks with longer trajectories.

Humans are able to describe image contents with coarse to fine details as they wish. However, most image captioning models are intention-agnostic which cannot generate diverse descriptions according to different user intentions initiatively. In this work, we propose the Abstract Scene Graph (ASG) structure to represent user intention in fine-grained level and control what and how detailed the generated description should be. The ASG is a directed graph consisting of three types of abstract nodes (object, attribute, relationship) grounded in the image without any concrete semantic labels. Thus it is easy to obtain either manually or automatically. From the ASG, we propose a novel ASG2Caption model, which is able to recognise user intentions and semantics in the graph, and therefore generate desired captions following the graph structure. Our model achieves better controllability conditioning on ASGs than carefully designed baselines on both VisualGenome and MSCOCO datasets. It also significantly improves the caption diversity via automatically sampling diverse ASGs as control signals. Code will be released at https://github.com/cshizhe/asg2cap.

Cross-modal retrieval between videos and texts has attracted growing attentions due to the rapid emergence of videos on the web. The current dominant approach is to learn a joint embedding space to measure cross-modal similarities. However, simple embeddings are insufficient to represent complicated visual and textual details, such as scenes, objects, actions and their compositions. To improve fine-grained video-text retrieval, we propose a Hierarchical Graph Reasoning (HGR) model, which decomposes video-text matching into global-to-local levels. The model disentangles text into a hierarchical semantic graph including three levels of events, actions, entities, and generates hierarchical textual embeddings via attention-based graph reasoning. Different levels of texts can guide the learning of diverse and hierarchical video representations for cross-modal matching to capture both global and local details. Experimental results on three video-text datasets demonstrate the advantages of our model. Such hierarchical decomposition also enables better generalization across datasets and improves the ability to distinguish fine-grained semantic differences. Code will be released at https: //github.com/cshizhe/hgr_v2t.

The human-object interaction detection (HOI) task aims to localize human and objects in an input image and predict their relationships, which is essential for understanding human behaviors in complex scenes. Due to the human-centric nature of the HOI task, it is beneficial to make use of human-related knowledge such as human skeletons to infer fine-grained human-object interactions. However, previous works simply embed skeletons via convolutional networks, which fail to capture structured connections in human skeletons and ignore the object influence. In this work, we propose a Skeleton-based Interactive Graph Network (SIGN) to capture fine-grained human-object interactions via encoding interactive graphs between keypoints in human skeletons and object from spatial and appearance aspects. Experimental results demonstrate the effectiveness of our SIGN model, which achieves significant improvement over baselines and outperforms other state-of-the-art methods on two benchmarks.

Most current image captioning systems focus on describing general image content, and lack background knowledge to deeply understand the image, such as exact named entities or concrete events. In this work, we focus on the entity-aware news image captioning task which aims to generate informative captions by leveraging the associated news articles to provide background knowledge about the target image. However, due to the length of news articles, previous works only employ news articles at the coarse article or sentence level, which are not fine-grained enough to refine relevant events and choose named entities accurately. To overcome these limitations, we propose an Information Concentrated Entity-aware news image CAPtioning (ICECAP) model, which progressively concentrates on relevant textual information within the corresponding news article from the sentence level to the word level. Our model first creates coarse concentration on relevant sentences using a cross-modality retrieval model and then generates captions by further concentrating on relevant words within the sentences. Extensive experiments on both BreakingNews and GoodNews datasets demonstrate the effectiveness of our proposed method, which outperforms other state-of-the-arts.

Bilingual lexicon induction, translating words from the source language to the target language, is a long-standing natural language processing task. Recent endeavors prove that it is promising to employ images as pivot to learn the lexicon induction without reliance on parallel corpora. However, these vision-based approaches simply associate words with entire images, which are constrained to translate concrete words and require object-centered images. We humans can understand words better when they are within a sentence with context. Therefore, in this paper, we propose to utilize images and their associated captions to address the limitations of previous approaches. We propose a multi-lingual caption model trained with different mono-lingual multimodal data to map words in different languages into joint spaces. Two types of word representation are induced from the multi-lingual caption model: linguistic features and localized visual features. The linguistic feature is learned from the sentence contexts with visual semantic constraints, which is beneficial to learn translation for words that are less visual-relevant. The localized visual feature is attended to the region in the image that correlates to the word, so that it alleviates the image restriction for salient visual representation. The two types of features are complementary for word translation. Experimental results on multiple language pairs demonstrate the effectiveness of our proposed method, which substantially outperforms previous vision-based approaches without using any parallel sentences or supervision of seed word pairs.

Weiying Wang, Yongcheng Wang, Shizhe Chen, Qin Jin. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

With the development of globalization, automatic emotion recognition has faced a new challenge in the multi-culture scenario - to generalize across different cultures. Previous works mainly rely on multi-cultural datasets to address the cross-culture discrepancy, which are expensive to collect. In this paper, we propose an adversarial learning framework to alleviate the culture influence on multimodal emotion recognition. We treat the emotion recognition and culture recognition as two adversarial tasks. The emotion feature embedding is trained to improve the emotion recognition but to confuse the culture recognition, so that it is more emotion-salient and culture-invariant for cross-culture emotion recognition. Our approach is applicable to both mono-culture and multi-culture emotion datasets. Extensive experiments demonstrate that the proposed method significantly outperforms previous baselines in both cross-culture and multi-culture evaluations.

The neural machine translation model has suffered from the lack of large-scale parallel corpora. In contrast, we humans can learn multi-lingual translations even without parallel texts by referring our languages to the external world. To mimic such human learning behavior, we employ images as pivots to enable zero-resource translation learning. However, a picture tells a thousand words, which makes multi-lingual sentences pivoted by the same image noisy as mutual translations and thus hinders the translation model learning. In this work, we propose a progressive learning approach for image-pivoted zero-resource machine translation. Since words are less diverse when grounded in the image, we first learn word-level translation with image pivots, and then progress to learn the sentence-level translation by utilizing the learned word translation to suppress noises in image-pivoted multi-lingual sentences. Experimental results on two widely used image-pivot translation datasets, IAPR-TC12 and Multi30k, show that the proposed approach significantly outperforms other state-of-the-art methods.

Generating image descriptions in different languages is essential to satisfy users worldwide. However, it is prohibitively expensive to collect large-scale paired image-caption dataset for every target language which is critical for training descent image captioning models. Previous works tackle the unpaired cross-lingual image captioning problem through a pivot language, which is with the help of paired image-caption data in the pivot language and pivot-to-target machine translation models. However, such language-pivoted approach suffers from inaccuracy brought by the pivot-to-target translation, including disfluency and visual irrelevancy errors. In this paper, we propose to generate cross-lingual image captions with self-supervised rewards in the reinforcement learning framework to alleviate these two types of errors. We employ self-supervision from mono-lingual corpus in the target language to provide fluency reward, and propose a multi-level visual semantic matching model to provide both sentence-level and concept-level visual relevancy rewards. We conduct extensive experiments for unpaired cross-lingual image captioning in both English and Chinese respectively on two widely used image caption corpora. The proposed approach achieves significant performance improvement over state-of-the-art methods.

A storyboard is a sequence of images to illustrate a story containing multiple sentences, which has been a key process to create different story products. In this paper, we tackle a new multimedia task of automatic storyboard creation to facilitate this process and inspire human artists. Inspired by the fact that our understanding of languages is based on our past experience, we propose a novel inspire-and-create framework with a story-to-image retriever that selects relevant cinematic images for inspiration and a storyboard creator that further refines and renders images to improve the relevancy and visual consistency. The proposed retriever dynamically employs contextual information in the story with hierarchical attentions and applies dense visual-semantic matching to accurately retrieve and ground images. The creator then employs three rendering steps to increase the flexibility of retrieved images, which include erasing irrelevant regions, unifying styles of images and substituting consistent characters. We carry out extensive experiments on both in-domain and out-of-domain visual story datasets. The proposed model achieves better quantitative performance than the state-of-the-art baselines for storyboard creation. Qualitative visualizations and user studies further verify that our approach can create high-quality storyboards even for stories in the wild.

In this paper, we present our solutions to the grand challenge task "Relation Understanding in Videos" in ACM Multimedia 2019. The challenge task aims to detect instances of target visual relations in a video, where a visual relation instance is represented by a relation triplet with the trajectories of the subject and object. It seems that it is similar to the image relation detection task with the input changed from images to videos. However, video relation detection requires a much more complex pipeline which not only needs to detect objects in each frame, but also track them in the temporal direction. In this challenge, we follow the basic pipeline structure which consists of 3 main separate components: an object detector, an object tracker and a relation predictor. Based on our analysis, there exist data unbalance and label missing problems in the VidOR dataset. We exploit two simple but effective methods to alleviate these problems. We also use trajectory Non-Maximum Suppression and a sliding window method to address the redundancy of trajectory proposals. Experimental results on the challenge task demonstrate that by applying these proposed approaches over the state-of-the-art relation prediction model, the video relation detection performance can be improved with the [email protected] and [email protected] reaching 0.3305 and 0.3507 respectively.

Visual relations, which describe various types of interactions between two objects in the image, can provide critical information for comprehensive semantic understanding of the image. Multiple cues related to the objects can contribute to visual relation detection, which mainly include appearances, spacial locations and semantic meanings. It is of great importance to represent different cues and combine them effectively for visual relation detection. However, in previous works, the appearance representation is simply realized by global visual representation based on the bounding boxes of objects, which may not capture salient regions of the interaction between two objects, and the different cue representations are equally concatenated without considering their different contributions for different relations. In this work, we propose a multi-level attention visual relation detection model (MLA-VRD), which generates salient appearance representation via a multi-stage appearance attention strategy and adaptively combine different cues with different importance weighting via a multi-cue attention strategy. Extensive experiment results on two widely used visual relation detection datasets, VRD and Visual Genome, demonstrate the effectiveness of our proposed model which significantly outperforms the previous state-of-the-arts. Our proposed model also achieves superior performance under the zero-shot learning condition, which is an important ordeal for testing the generalization ability of visual relation detection models.

Automatic video description generation (a.k.a video captioning) is one of the ultimate goals for video understanding. Despite the wide range of applications such as video indexing and retrieval etc., the video captioning task remains quite challenging due to the complexity and diversity of video content. First, open-domain videos cover a broad range of topics, which results in highly variable vocabularies and expression styles to describe the video contents. Second, videos naturally contain multiple modalities including image, motion, and acoustic media. The information provided by different modalities differs in different conditions. In this paper, we propose a novel topic-guided video captioning model to address the above-mentioned challenges in video captioning. Our model consists of two joint tasks, namely, latent topic generation and topic-guided caption generation. The topic generation task aims to automatically predict the latent topic of the video. Since there is no groundtruth topic information, we mine multimodal topics in an unsupervised fashion based on video contents and annotated captions, and then distill the topic distribution to a topic prediction model. In the topic-guided generation task, we employ the topic guidance for two purposes. The first is to narrow down the language complexity across topics, where we propose the topic-aware decoder to leverage the latent topics to induce topic-related language models. The decoder is also generic and can be integrated with a temporal attention mechanism. The second is to dynamically attend to important modalities by topics, where we propose a flexible topic-guided multimodal ensemble framework and use the topic gating network to determine the attention weights. The two tasks are correlated with each other, and they collaborate to generate more detailed and accurate video captions. Our extensive experiments on two public benchmark datasets MSR-VTT and Youtube2Text demonstrate the effectiveness of the proposed topic-guided video captioning system, which achieves state-of-the-art performance on both datasets.

Audio event recognition (AER) has been an important research problem with a wide range of applications. However, it is very challenging to develop large scale audio event recognition models. On the one hand, usually there are only "weak" labeled audio training data available, which only contains labels of audio events without temporal boundaries. On the other hand, the distribution of audio events is generally long-tailed, with only a few positive samples for large amounts of audio events. These two issues make it hard to learn discriminative acoustic features to recognize audio events especially for long-tailed events. In this paper, we propose a novel class-aware self-attention mechanism with attention factor sharing to generate discriminative clip-level features for audio event recognition. Since a target audio event only occurs in part of an entire audio clip and its corresponding temporal interval varies, the proposed class-aware self-attention approach learns to highlight relevant temporal intervals and to suppress irrelevant noises at the same time. In order to learn attention patterns effectively for those long-tailed events, we combine both the domain knowledge and data driven strategies to share attention factors in the proposed attention mechanism, which transfers the common knowledge learned from other similar events to the rare events. The proposed attention mechanism is a pluggable component and can be trained end-to-end in the overall AER model. We evaluate our model on a large-scale audio event corpus "Audio Set" with both short-term and long-term acoustic features. The experimental results demonstrate the effectiveness of our model, which improves the overall audio event recognition performance with different acoustic features especially for events with low resources. Moreover, the experiments also show that our proposed model is able to learn new audio events with a few training examples effectively and efficiently without disturbing the previously learned audio events.

In this paper, we present our methods for the facial AU occurrence subtask in the FG 2017 Facial Expression Recognition and Analysis challenge (FERA 2017). The task is to detect 10 AUs that occur frequently in the BP4D database. In our approach, we utilized the hand-crafted features (LBPTOP) and learned features (CNN features) which encode the temporal dynamics and static facial appearance respectively. Multiple classifiers including SVM and Random Forest are used to recognize the AU occurrence. Among all different features, the learned CNN features are the most discriminative feature for the task. And the combinations of the multiple features are beneficial. To capture the relationships among AUs, we also apply multi-AUs late fusion, which further boost the performance. Our system achieves an average macro F1 of 49.8% and an average accuracy of 69.4% in the testing set which outperforms the baseline (macro F1 45.2% and average accuracy 56.1%).

This paper presents our methods to the Audio-Video Based Emotion Recognition subtask in the 2017 Emotion Recognition in the Wild (EmotiW) Challenge. The task aims to predict one of the seven basic emotions for short video segments. We extract different features from audio and facial expression modalities. We also explore the temporal LSTM model with the input of frame facial features, which improves the performance of the non-temporal model. The fusion of different modality features and the temporal model lead us to achieve a 58.5% accuracy on the testing set, which shows the effectiveness of our methods.

Generating video descriptions in natural language (a.k.a. video captioning) is a more challenging task than image captioning as the videos are intrinsically more complicated than images in two aspects. First, videos cover a broader range of topics, such as news, music, sports and so on. Second, multiple topics could coexist in the same video. In this paper, we propose a novel caption model, topic-guided model (TGM), to generate topic-oriented descriptions for videos in the wild via exploiting topic information. In addition to predefined topics, i.e., category tags crawled from the web, we also mine topics in a data-driven way based on training captions by an unsupervised topic mining model. We show that data-driven topics reflect a better topic schema than the predefined topics. As for testing video topic prediction, we treat the topic mining model as teacher to train the student, the topic prediction model, by utilizing the full multi-modalities in the video especially the speech modality. We propose a series of caption models to exploit topic guidance, including implicitly using the topics as input features to generate words related to the topic and explicitly modifying the weights in the decoder with topics to function as an ensemble of topic-aware language decoders. Our comprehensive experimental results on the current largest video caption dataset MSR-VTT prove the effectiveness of our topic-guided model, which significantly surpasses the winning performance in the 2016 MSR video to language challenge.

The topic diversity of open-domain videos leads to various vocabularies and linguistic expressions in describing video contents, and therefore, makes the video captioning task even more challenging. In this paper, we propose an unified caption framework, M&M TGM, which mines multimodal topics in unsupervised fashion from data and guides the caption decoder with these topics. Compared to pre-defined topics, the mined multimodal topics are more semantically and visually coherent and can reflect the topic distribution of videos better. We formulate the topic-aware caption generation as a multi-task learning problem, in which we add a parallel task, topic prediction, in addition to the caption task. For the topic prediction task, we use the mined topics as the teacher to train a student topic prediction model, which learns to predict the latent topics from multimodal contents of videos. The topic prediction provides intermediate supervision to the learning process. As for the caption task, we propose a novel topic-aware decoder to generate more accurate and detailed video descriptions with the guidance from latent topics. The entire learning procedure is end-to-end and it optimizes both tasks simultaneously. The results from extensive experiments conducted on the MSR-VTT and Youtube2Text datasets demonstrate the effectiveness of our proposed model. M&M TGM not only outperforms prior state-of-the-art methods on multiple evaluation metrics and on both benchmark datasets, but also achieves better generalization ability.

Generating natural language descriptions for videos (a.k.a video captioning) has attracted much research attention in recent years, and a lot of models have been proposed to improve the caption performance. However, due to the rapid progress in dataset expansion and feature representation, newly proposed caption models have been evaluated on different settings, which makes it unclear about the contributions from either features or models. Therefore, in this work we aim to gain a deep understanding about "where are we" for the current development of video captioning. First, we carry out extensive experiments to identify the contribution from different components in video captioning task and make fair comparison among several state-of-the-art video caption models. Second, we discover that these state-of-the-art models are complementary so that we could benefit from "wisdom of the crowd" through ensembling and reranking. Finally, we give a preliminary answer to the question "how far are we from the human-level performance in general'' via a series of carefully designed experiments. In summary, our caption models achieve the state-of-the-art performance on the MSR-VTT 2017 challenge, and it is comparable with the average human-level performance on current caption metrics. However, our analysis also shows that we still have a long way to go, such as further improving the generalization ability of current caption models.

In this paper, we present our methods to the Audio-Video Based Emotion Recognition subtask in the 2016 Emotion Recognition in the Wild (EmotiW) Challenge. The task is to predict one of the seven basic emotions for the characters in the video clips extracted from movies or TV shows. In our approach, we explore various multimodal features from audio, facial image and video motion modalities. The audio features contain statistical acoustic features, MFCC Bag-of-Audio-Words and MFCC Fisher Vectors. For image related features, we extract hand-crafted features (LBP-TOP and SPM Dense SIFT) and learned features (CNN features). The improved Dense Trajectory is used as the motion related features. We train SVM, Random Forest and Logistic Regression classifiers for each kind of feature. Among them, MFCC fisher vector is the best acoustic features and the facial CNN feature is the most discriminative feature for emotion recognition. We utilize late fusion to combine different modality features and achieve a 50.76% accuracy on the testing set, which significantly outperforms the baseline test accuracy of 40.47%.

Continuous dimensional emotion prediction is a challenging task where the fusion of various modalities usually achieves state-of-the-art performance such as early fusion or late fusion. In this paper, we propose a novel multi-modal fusion strategy named conditional attention fusion, which can dynamically pay attention to different modalities at each time step. Long-short term memory recurrent neural networks (LSTM-RNN) is applied as the basic uni-modality model to capture long time dependencies. The weights assigned to different modalities are automatically decided by the current input features and recent history information rather than being fixed at any kinds of situation. Our experimental results on a benchmark dataset AVEC2015 show the effectiveness of our method which outperforms several common fusion strategies for valence prediction.

Describing videos with natural language is one of the ultimate goals of video understanding. Video records multi-modal information including image, motion, aural, speech and so on. MSR Video to Language Challenge provides a good chance to study multi-modality fusion in caption task. In this paper, we propose the multi-modal fusion encoder and integrate it with text sequence decoder into an end-to-end video caption framework. Features from visual, aural, speech and meta modalities are fused together to represent the video contents. Long Short-Term Memory Recurrent Neural Networks (LSTM-RNNs) are then used as the decoder to generate natural language sentences. Experimental results show the effectiveness of multi-modal fusion encoder trained in the end-to-end framework, which achieved top performance in both common metrics evaluation and human evaluation.

In this paper we explore one of the key aspects in building an emotion recognition system: generating suitable feature representations. We generate feature representations from both acoustic and lexical levels. At the acoustic level, we first extract low-level features such as intensity, F0, jitter, shimmer and spectral contours etc. We then generate different acoustic feature representations based on these low-level features, including statistics over these features, a new representation derived from a set of low-level acoustic codewords, and a new representation from Gaussian Supervectors. At the lexical level, we propose a new feature representation named emotion vector (eVector). We also use the traditional Bag-of-Words (BoW) feature. We apply these feature representations for emotion recognition and compare their performance on the USC-IEMOCAP database. We also combine these different feature representations via early fusion and late fusion. Our experimental results show that late fusion of both acoustic and lexical features achieves four-class emotion recognition accuracy of 69.2%.