We introduce VideoComp, a benchmark and learning framework for advancing video-text compositionality understanding, aimed at improving vision-language models (VLMs) in fine-grained temporal alignment. Unlike existing benchmarks focused on static image-text compositionality or isolated single-event videos, our benchmark targets alignment in continuous multi-event videos. Leveraging video-text datasets with temporally localized event captions (e.g. ActivityNet-Captions, YouCook2), we construct two compositional benchmarks, ActivityNet-Comp and YouCook2-Comp. We create challenging negative samples with subtle temporal disruptions such as reordering, action word replacement, partial captioning, and combined disruptions. These benchmarks comprehensively test models’ compositional sensitivity across extended, cohesive video-text sequences. To improve model performance, we propose a hierarchical pairwise preference loss that strengthens alignment with temporally accurate pairs and gradually penalizes increasingly disrupted ones, encouraging fine-grained compositional learning. To mitigate the limited availability of densely annotated video data, we introduce a pretraining strategy that concatenates short video-caption pairs to simulate multi-event sequences. We evaluate video-text foundational models and large multimodal models (LMMs) on our benchmark, identifying both strengths and areas for improvement in compositionality. Overall, our work provides a comprehensive framework for evaluating and enhancing model capabilities in achieving fine-grained, temporally coherent video-text alignment.

Visual grounding tasks aim to localize image regions based on natural language references. In this work, we ex-plore whether generative VLMs predominantly trained on image-text data could be leveraged to scale up the text an-notation of visual grounding data. We find that grounding knowledge already exists in generative VLM and can be elicited by proper prompting. We thus prompt a VLM to generate object-level descriptions by feeding it object regions from existing object detection datasets. We fur-ther propose attribute modeling to explicitly capture the im-portant object attributes, and spatial relation modeling to capture inter-object relationship, both of which are common linguistic pattern in referring expression. Our constructed dataset (500K images, 1M objects, 16M referring expressions) is one of the largest grounding datasets to date, and the first grounding dataset with purely model-generated queries and human-annotated objects. To verify the qual-ity of this data, we conduct zero-shot transfer experiments to the popular RefCoco benchmarks for both referring expression comprehension (REC) and segmentation (RES) tasks. On both tasks, our model significantly outperform the state-of-the-art approaches without using human anno-tated visual grounding data. Our results demonstrate the promise of generative VLM to scale up visual grounding in the real world.

One of the main challenges of multimodal learning is combining multiple heterogeneous modalities, e.g., video, audio, and text. Video and audio are obtained at much higher rates than text and are roughly aligned in time. They are often not synchronized with text, which comes as a global context, e.g. a title, or a description. Furthermore, video and audio inputs are of much larger volumes, and grow as the video length increases, which naturally requires more compute dedicated to these modalities, and makes modeling of long-range dependencies harder. We here decouple the multimodal modeling, dividing it into separate autoregressive models, processing the inputs according to the characteristics of the modalities. We propose a multimodal model, consisting of an autoregressive component for the time-synchronized modalities (audio and video), and an autoregressive component for the context modalities which are not necessarily aligned in time but are still sequential. To address the long-sequences of the video-audio inputs, we further partition the video and audio sequences in consecutive snippets and autoregressively process their representations. To that end, we propose a Combiner mechanism, which models the audio-video information jointly, producing compact but expressive representations. This allows us to scale to 512 input video frames without increase in model parameters. Our approach achieves the state-of-the-art on multiple well established multimodal benchmarks. It effectively addresses the high computational demand of media inputs by learning compact representations, controlling the sequence length of the audio-video feature representations, and modeling their dependencies in time.

Video Panoptic Segmentation (VPS) aims to achieve comprehensive pixel-level scene understanding by segmenting all pixels and associating objects in a video. Current solutions can be categorized into online and near-online approaches. Evolving over the time, each category has its own specialized designs, making it nontrivial to adapt models between different categories. To alleviate the discrepancy, in this work, we propose a unified approach for online and near-online VPS. The meta architecture of the proposed Video-kMaX consists of two components: within-clip segmenter (for clip-level segmentation) and cross-clip associater (for association beyond clips). We propose clip-kMaX (clip k-means mask transformer) and LA-MB (location-aware memory buffer) to instantiate the segmenter and associater, respectively. Our general formulation includes the online scenario as a special case by adopting clip length of one. Without bells and whistles, Video-kMaX sets a new state-of-the-art on KITTI-STEP and VIPSeg for video panoptic segmentation Code and models are available at this link.

This paper proposes a textless training method for many-to-many multilingual speech-to-speech translation that can also benefit the transfer of pre-trained knowledge to text-based systems, text-to-speech synthesis and text-to-speech translation. To this end, we represent multilingual speech with speech units that are the discretized representations of speech features derived from a self-supervised speech model. By treating the speech units as pseudo-text, we can focus on the linguistic content of the speech, which can be easily associated with both speech and text modalities at the phonetic level information. By setting both the inputs and outputs of our learning problem as speech units, we propose to train an encoder-decoder model in a many-to-many spoken language translation setting, namely Unit-to-Unit Translation (UTUT). Specifically, the encoder is conditioned on the source language token to correctly understand the input spoken language, while the decoder is conditioned on the target language token to generate the translated speech in the target language. Therefore, during the training, the model can build the knowledge of how languages are comprehended and how to relate them to different languages. Since speech units can be easily associated from both audio and text by quantization and phonemization respectively, the trained model can easily transferred to text-related tasks, even if it is trained in a textless manner. We demonstrate that the proposed UTUT model can be effectively utilized not only for Speech-to-Speech Translation (S2ST) but also for multilingual Text-to-Speech Synthesis (T2S) and Text-to-Speech Translation (T2ST), requiring only minimal fine-tuning steps on text inputs. By conducting comprehensive experiments encompassing various languages, we validate the efficacy of the proposed method across diverse multilingual tasks. Moreover, thanks to the many-to-many language training, we show that the UTUT can also perform language translations for novel language pairs that are not present during training as pairs, which has not well been explored in the previous literature.

We present Region-aware Open-vocabulary Vision Transformers (RO-ViT) - a contrastive image-text pretraining recipe to bridge the gap between image-level pretraining and open-vocabulary object detection. At the pretraining phase, we propose to randomly crop and resize regions of positional embeddings instead of using the whole image positional embeddings. This better matches the use of positional embeddings at region-level in the detection finetuning phase. In addition, we replace the common softmax cross entropy loss in contrastive learning with focal loss to better learn the informative yet difficult examples. Finally, we leverage recent advances in novel object proposals to improve open-vocabulary detection finetuning. We evaluate our full model on the LVIS and COCO open-vocabulary detection benchmarks and zero-shot transfer. RO-ViT achieves a state-of-the-art 32.1 APron LVIS, surpassing the best existing approach by +5.8 points in addition to competitive zero-shot transfer detection. Surprisingly, RO-ViT improves the image-level representation as well and achieves the state of the art on 9 out of 12 metrics on COCO and Flickr image-text retrieval benchmarks, outperforming competitive approaches with larger models.

We present Contrastive Feature Masking Vision Trans-former (CFM-ViT) - an image-text pretraining methodology that achieves simultaneous learning of image- and region-level representation for open-vocabulary object detection (OVD). Our approach combines the masked autoencoder (MAE) objective into the contrastive learning objective to improve the representation for localization tasks. Un-like standard MAE, we perform reconstruction in the joint image-text embedding space, rather than the pixel space as is customary with the classical MAE method, which causes the model to better learn region-level semantics. Moreover, we introduce Positional Embedding Dropout (PED) to address scale variation between image-text pretraining and detection finetuning by randomly dropping out the positional embeddings during pretraining. PED improves detection performance and enables the use of a frozen ViT backbone as a region classifier, preventing the forgetting of open-vocabulary knowledge during detection finetuning. On LVIS open-vocabulary detection benchmark, CFM-ViT achieves a state-of-the-art 33.9 APr, surpassing the best approach by 7.6 points and achieves better zero-shot detection transfer. Finally, CFM-ViT acquires strong image-level representation, outperforming the state of the art on 8 out of 12 metrics on zero-shot image-text retrieval benchmarks.

Energy-aware computing is one of the most critical issues in modern computing systems. Linux has introduced the schedutil governor since its 4.7 kernel release, which dynamically scales the processor frequency level to reduce energy consumption. Although it has been widely used in most Linux-based systems, the governor often makes inaccurate decisions in frequency selection, thus leading to unnecessary energy consumption. In this paper, we propose an enhanced governor as an alternative. We first rigorously analyze the schedutil's policy and then find out that it does not take into account memory stalls when characterizing CPU performance via CPI (cycles per instruction). This yields inherent inaccuracy, particularly in modern SoCs where multiple CPU cores and accelerators incur a huge amount of memory traffic over the system bus. To rectify this problem, we reformulate the CPU performance estimation function of the schedutil governor via the memory stall cycle ratio so that it can dynamically reflect the effects of changes in the processor's frequency and the system's memory contention. We show that our CPU performance estimation function is easily integrated into schedutil. We also show that the memory stall cycle ratio, the key element of the function, can be efficiently calculated at runtime with performance monitoring units (PMU) commonly available in most modern SoCs. We have implemented our governor and conducted extensive experiments to validate its effectiveness. Experimental results show that our governor saves more energy than schedutil by up to 28.91% without noticeable performance degradation.

We present TubeFormer-DeepLab, the first attempt to tackle multiple core video segmentation tasks in a unified manner. Different video segmentation tasks (e.g., video semantic/instance/panoptic segmentation) are usually considered as distinct problems. State-of-the-art models adopted in the separate communities have diverged, and radically different approaches dominate in each task. By contrast, we make a crucial observation that video segmentation tasks could be generally formulated as the problem of assigning different predicted labels to video tubes (where a tube is obtained by linking segmentation masks along the time axis) and the labels may encode different values depending on the target task. The observation motivates us to develop TubeFormer-DeepLab, a simple and effective video mask transformer model that is widely applicable to multiple video segmentation tasks. TubeFormer-DeepLab directly predicts video tubes with task-specific labels (either pure semantic categories, or both semantic categories and instance identities), which not only significantly simplifies video segmentation models, but also advances state-of-the-art results on multiple video segmentation benchmarks.

We propose Clustering Mask Transformer (CMT-DeepLab), a transformer-based framework for panoptic segmentation designed around clustering. It rethinks the existing transformer architectures used in segmentation and detection; CMT-DeepLab considers the object queries as cluster centers, which fill the role of grouping the pixels when applied to segmentation. The clustering is computed with an alternating procedure, by first assigning pixels to the clusters by their feature affinity, and then updating the cluster centers and pixel features. Together, these operations comprise the Clustering Mask Transformer (CMT) layer, which produces cross-attention that is denser and more consistent with the final segmentation task. CMT-DeepLab improves the performance over prior art significantly by 4.4% PQ, achieving a new state-of-the-art of 55.7% PQ on the COCO test-dev set.

Fashion attribute editing aims to manipulate fashion images based on a user-specified attribute, while preserving the details of the original image as intact as possible. Recent works in this domain have mainly focused on direct manipulation of the raw RGB pixels, which only allows to perform edits involving relatively small shape changes (e.g., sleeves). The goal of our Virtual Personal Tailoring Network (VPTNet) is to extend the editing capabilities to much larger shape changes of fashion items, such as cloth length. To achieve this goal, we decouple the fashion attribute editing task into two conditional stages: shape-then-appearance editing. To this aim, we propose a shape editing network that employs a semantic parsing of the fashion image as an interface for manipulation. Compared to operating on the raw RGB image, our parsing map editing enables performing more complex shape editing operations. Second, we introduce an appearance completion network that takes the previous stage results and completes the shape difference regions to produce the final RGB image. Qualitative and quantitative experiments on the DeepFashion-Synthesis dataset confirm that VPTNet outperforms state-of-the-art methods for both small and large shape attribute editing.

A holistic understanding of dynamic scenes is of fundamental importance in real-world computer vision problems such as autonomous driving, augmented reality and spatio-temporal reasoning. In this paper, we propose a new computer vision benchmark: Video Panoptic Segmentation (VPS). To study this important problem, we present two datasets, Cityscapes-VPS and VIPER together with a new evaluation metric, video panoptic quality (VPQ). We also propose VPSNet++, an advanced video panoptic segmentation network, which simultaneously performs classification, detection, segmentation, and tracking of all identities in videos. Specifically, VPSNet++ builds upon a top-down panoptic segmentation network by adding pixel-level feature fusion head and object-level association head. The former temporally augments the pixel features while the latter performs object tracking. Furthermore, we propose panoptic boundary learning as an auxiliary task, and instance discrimination learning which learns spatio-temporally clustered pixel embedding for individual thing or stuff regions, i.e., exactly the objective of the video panoptic segmentation problem. Our VPSNet++ significantly outperforms the default VPSNet, i.e., FuseTrack baseline, and achieves state-of-the-art results on both Cityscapes-VPS and VIPER datasets. The datasets, metric, and models are publicly available at https://github.com/mcahny/vps.

Temporal correspondence - linking pixels or objects across frames - is a fundamental supervisory signal for the video models. For the panoptic understanding of dynamic scenes, we further extend this concept to every segment. Specifically, we aim to learn coarse segment-level matching and fine pixel-level matching together. We implement this idea by designing two novel learning objectives. To validate our proposals, we adopt a deep siamese model and train the model to learn the temporal correspondence on two different levels (i.e., segment and pixel) along with the target task. At inference time, the model processes each frame independently without any extra computation and post-processing. We show that our per-frame inference model can achieve new state-of-the-art results on Cityscapes-VPS and VIPER datasets. Moreover, due to its high efficiency, the model runs in a fraction of time (3×) compared to the previous state-of-the-art approach.

In this paper, we aim at synthesizing a free-viewpoint video of an arbitrary human performance using sparse multi-view cameras. Recently, several works have addressed this problem by learning person-specific neural radiance fields (NeRF) to capture the appearance of a particular human. In parallel, some work proposed to use pixel-aligned features to generalize radiance fields to arbitrary new scenes and objects. Adopting such generalization approaches to humans, however, is highly challenging due to the heavy occlusions and dynamic articulations of body parts. To tackle this, we propose Neural Human Performer, a novel approach that learns generalizable neural radiance fields based on a parametric human body model for robust performance capture. Specifically, we first introduce a temporal transformer that aggregates tracked visual features based on the skeletal body motion over time. Moreover, a multi-view transformer is proposed to perform cross-attention between the temporally-fused features and the pixel-aligned features at each time step to integrate observations on the fly from multiple views. Experiments on the ZJU-MoCap and AIST datasets show that our method significantly outperforms recent generalizable NeRF methods on unseen identities and poses.

Pursuing a more coherent scene understanding towards real-time vision applications, single-stage instance segmentation has recently gained popularity, achieving a simpler and more efficient design than its two-stage counterparts. Besides, its global mask representation often leads to superior accuracy to the two-stage Mask R-CNN which has been dominant thus far. Despite the promising advances in single-stage methods, finer delineation of instance boundaries still remains unexcavated. Indeed, boundary information provides a strong shape representation that can operate in synergy with the fully-convolutional mask features of the single-stage segmenter. In this work, we propose Boundary Basis based Instance Segmentation(B2Inst) to learn a global boundary representation that can complement existing global-mask-based methods that are often lacking high-frequency details. Besides, we devise a unified quality measure of both mask and boundary and introduce a network block that learns to score the per-instance predictions of itself. When applied to the strongest baselines in single-stage instance segmentation, our B2Inst leads to consistent improvements and accurately parse out the instance boundaries in a scene. Regardless of being single-stage or two-stage frameworks, we outperform the existing state-of-the-art methods on the COCO dataset with the same ResNet-50 and ResNet-101 backbones.

Visual storytelling is a task of creating a short story based on photo streams. Unlike existing visual captioning, storytelling aims to contain not only factual descriptions, but also human-like narration and semantics. However, the VIST dataset consists only of a small, fixed number of photos per story. Therefore, the main challenge of visual storytelling is to fill in the visual gap between photos with narrative and imaginative story. In this paper, we propose to explicitly learn to imagine a storyline that bridges the visual gap. During training, one or more photos is randomly omitted from the input stack, and we train the network to produce a full plausible story even with missing photo(s). Furthermore, we propose for visual storytelling a hide-and-tell model, which is designed to learn non-local relations across the photo streams and to refine and improve conventional RNN-based models. In experiments, we show that our scheme of hide-and-tell, and the network design are indeed effective at storytelling, and that our model outperforms previous state-of-the-art methods in automatic metrics. Finally, we qualitatively show the learned ability to interpolate storyline over visual gaps.

Panoptic segmentation has become a new standard of visual recognition task by unifying previous semantic segmentation and instance segmentation tasks in concert. In this paper, we propose and explore a new video extension of this task, called video panoptic segmentation. The task requires generating consistent panoptic segmentation as well as an association of instance ids across video frames. To invigorate research on this new task, we present two types of video panoptic datasets. The first is a re-organization of the synthetic VIPER dataset into the video panoptic format to exploit its large-scale pixel annotations. The second is a temporal extension on the Cityscapes val. set, by providing new video panoptic annotations (Cityscapes-VPS). Moreover, we propose a novel video panoptic segmentation network (VPSNet) which jointly predicts object classes, bounding boxes, masks, instance id tracking, and semantic segmentation in video frames. To provide appropriate metrics for this task, we propose a video panoptic quality (VPQ) metric and evaluate our method and several other baselines. Experimental results demonstrate the effectiveness of the presented two datasets. We achieve state-of-the-art results in image PQ on Cityscapes and also in VPQ on Cityscapes-VPS and VIPER datasets.

Human novel view synthesis aims to synthesize target views of a human subject given input images taken from one or more reference viewpoints. Despite significant advances in model-free novel view synthesis, existing methods present two major limitations when applied to complex shapes like humans. First, these methods mainly focus on simple and symmetric objects, e.g., cars and chairs, limiting their performances to fine-grained and asymmetric shapes. Second, existing methods cannot guarantee visual consistency across different adjacent views of the same object. To solve these problems, we present in this paper a learning framework for the novel view synthesis of human subjects, which explicitly enforces consistency across different generated views of the subject. Specifically, we introduce a novel multi-view supervision and an explicit rotational loss during the learning process, enabling the model to preserve detailed body parts and to achieve consistency between adjacent synthesized views. To show the superior performance of our approach, we present qualitative and quantitative results on the Multi-View Human Action (MVHA) dataset we collected (consisting of 3D human models animated with different Mocap sequences and captured from 54 different viewpoints), the Pose-Varying Human Model (PVHM) dataset, and ShapeNet. The qualitative and quantitative results demonstrate that our approach outperforms the state-of-the-art baselines in both per-view synthesis quality, and in preserving rotational consistency and complex shapes (e.g. fine-grained details, challenging poses) across multiple adjacent views in a variety of scenarios, for both humans and rigid objects.

Video inpainting aims to fill in spatio-temporal holes in videos with plausible content. Despite tremendous progress on deep learning-based inpainting of a single image, it is still challenging to extend these methods to video domain due to the additional time dimension. In this paper, we propose a recurrent temporal aggregation framework for fast deep video inpainting. In particular, we construct an encoder-decoder model, where the encoder takes multiple reference frames which can provide visible pixels revealed from the scene dynamics. These hints are aggregated and fed into the decoder. We apply a recurrent feedback in an auto-regressive manner to enforce temporal consistency in the video results. We propose two architectural designs based on this framework. Our first model is a blind video decaptioning network (BVDNet) that is designed to automatically remove and inpaint text overlays in videos without any mask information. Our BVDNet wins the first place in the ECCV Chalearn 2018 LAP Inpainting Competition Track 2: Video Decaptioning. Second, we propose a network for more general video inpainting (VINet) to deal with more arbitrary and larger holes. Video results demonstrate the advantage of our framework compared to state-of-the-art methods both qualitatively and quantitatively. The codes are available at https://github.com/mcahny/Deep-Video-Inpainting, and https://github.com/shwoo93/video_decaptioning.

In this paper, we address the problem of unsupervised video summarization that automatically extracts key-shots from an input video. Specifically, we tackle two critical issues based on our empirical observations: (i) Ineffective feature learning due to flat distributions of output importance scores for each frame, and (ii) training difficulty when dealing with longlength video inputs. To alleviate the first problem, we propose a simple yet effective regularization loss term called variance loss. The proposed variance loss allows a network to predict output scores for each frame with high discrepancy which enables effective feature learning and significantly improves model performance. For the second problem, we design a novel two-stream network named Chunk and Stride Network (CSNet) that utilizes local (chunk) and global (stride) temporal view on the video features. Our CSNet gives better summarization results for long-length videos compared to the existing methods. In addition, we introduce an attention mechanism to handle the dynamic information in videos. We demonstrate the effectiveness of the proposed methods by conducting extensive ablation studies and show that our final model achieves new state-of-the-art results on two benchmark datasets.

Self-supervised tasks such as colorization, inpainting and zigsaw puzzle have been utilized for visual representation learning for still images, when the number of labeled images is limited or absent at all. Recently, this worthwhile stream of study extends to video domain where the cost of human labeling is even more expensive. However, the most of existing methods are still based on 2D CNN architectures that can not directly capture spatio-temporal information for video applications. In this paper, we introduce a new self-supervised task called as Space-Time Cubic Puzzles to train 3D CNNs using large scale video dataset. This task requires a network to arrange permuted 3D spatio-temporal crops. By completing Space-Time Cubic Puzzles, the network learns both spatial appearance and temporal relation of video frames, which is our final goal. In experiments, we demonstrate that our learned 3D representation is well transferred to action recognition tasks, and outperforms state-of-the-art 2D CNN-based competitors on UCF101 and HMDB51 datasets.

Blind video decaptioning is a problem of automatically removing text overlays and inpainting the occluded parts in videos without any input masks. While recent deep learning based inpainting methods deal with a single image and mostly assume that the positions of the corrupted pixels are known, we aim at automatic text removal in video sequences without mask information. In this paper, we propose a simple yet effective framework for fast blind video decaptioning. We construct an encoder-decoder model, where the encoder takes multiple source frames that can provide visible pixels revealed from the scene dynamics. These hints are aggregated and fed into the decoder. We apply a residual connection from the input frame to the decoder output to enforce our network to focus on the corrupted regions only. Our proposed model was ranked in the first place in the ECCV Chalearn 2018 LAP Inpainting Competition Track2: Video decaptioning. In addition, we further improve this strong model by applying a recurrent feedback. The recurrent feedback not only enforces temporal coherence but also provides strong clues on where the corrupted pixels are. Both qualitative and quantitative experiments demonstrate that our full model produces accurate and temporally consistent video results in real time (50+ fps).

Video inpainting aims to fill spatio-temporal holes with plausible content in a video. Despite tremendous progress of deep neural networks for image inpainting, it is challenging to extend these methods to the video domain due to the additional time dimension. In this work, we propose a novel deep network architecture for fast video inpainting. Built upon an image-based encoder-decoder model, our framework is designed to collect and refine information from neighbor frames and synthesize still-unknown regions. At the same time, the output is enforced to be temporally consistent by a recurrent feedback and a temporal memory module. Compared with the state-of-the-art image inpainting algorithm, our method produces videos that are much more semantically correct and temporally smooth. In contrast to the prior video completion method which relies on time-consuming optimization, our method runs in near real-time while generating competitive video results. Finally, we applied our framework to video retargeting task, and obtain visually pleasing results.

As new display technologies (i.e. foldable phone and modular display) with variable aspect ratios emerge, content-aware video retargeting has attracted much attention from both academia and industry. The content-aware video retargeting aims to adjust the aspect ratio of a video sequence while preserving both, its content and its spatio-temporal consistency. This is a particularly challenging task since these two properties may drastically differ and contradict depending on the video characteristics. In this paper, we explore this conflict in the context of video retargeting, then we propose an appropriate solution to alleviate this issue using a deep recurrent convolutional neural network architecture. First of all, we present a method to generate multiple ground-truth labels under various aspect ratios. Using this dataset, our network is trained to predict various retargeted video candidates from a single input sequence. The resulting candidates present different properties, some of them with more emphasis on the content preservation while the others focus on the spatio-temporal consistency. Among the generated candidates, the final result which satisfy the best compromise is selected. A large set of qualitative and quantitative experiments shows the ability of our method for the content-aware video retargeting.

In this paper, we investigate the problem of unpaired video-to-video translation. Given a video in the source domain, we aim to learn the conditional distribution of the corresponding video in the target domain, without seeing any pairs of corresponding videos. While significant progress has been made in the unpaired translation of images, directly applying these methods to an input video leads to low visual quality due to the additional time dimension. In particular, previous methods suffer from semantic inconsistency (i.e., semantic label flipping) and temporal flickering artifacts. To alleviate these issues, we propose a new framework that is composed of carefully-designed generators and discriminators, coupled with two core objective functions: 1) content preserving loss and 2) temporal consistency loss. Extensive qualitative and quantitative evaluations demonstrate the superior performance of the proposed method against previous approaches. We further apply our framework to a domain adaptation task and achieve favorable results.

Objects and their relationships are critical contents for image understanding. A scene graph provides a structured description that captures these properties of an image. However, reasoning about the relationships between objects is very challenging and only a few recent works have attempted to solve the problem of generating a scene graph from an image. In this paper, we present a novel method that improves scene graph generation by explicitly modeling inter-dependency among the entire object instances. We design a simple and effective relational embedding module that enables our model to jointly represent connections among all related objects, rather than focus on an object in isolation. Our novel method significantly benefits two main parts of the scene graph generation task: object classification and relationship classification. Using it on top of a basic Faster R-CNN, our model achieves state-of-the-art results on the Visual Genome benchmark. We further push the performance by introducing global context encoding module and geometrical layout encoding module. We validate our final model, LinkNet, through extensive ablation studies, demonstrating its efficacy in scene graph generation.

In this paper, we explore methods of complicating selfsupervised tasks for representation learning. That is, we do severe damage to data and encourage a network to recover them. First, we complicate each of three powerful self-supervised task candidates: jigsaw puzzle, inpainting, and colorization. In addition, we introduce a novel complicated self-supervised task called "Completing damaged jigsaw puzzles" which is puzzles with one piece missing and the other pieces without color. We train a convolutional neural network not only to solve the puzzles, but also generate the missing content and colorize the puzzles. The recovery of the aforementioned damage pushes the network to obtain robust and general-purpose representations. We demonstrate that complicating the self-supervised tasks improves their original versions and that our final task learns more robust and transferable representations compared to the previous methods, as well as the simple combination of our candidate tasks. Our approach achieves state-of-the-art performance in transfer learning on PASCAL classification and semantic segmentation.

Weakly supervised semantic segmentation and localization have a problem of focusing only on the most important parts of an image since they use only image-level annotations. In this paper, we solve this problem fundamentally via two-phase learning. Our networks are trained in two steps. In the first step, a conventional fully convolutional network (FCN) is trained to find the most discriminative parts of an image. In the second step, the activations on the most salient parts are suppressed by inference conditional feedback, and then the second learning is performed to find the area of the next most important parts. By combining the activations of both phases, the entire portion of the target object can be captured. Our proposed training scheme is novel and can be utilized in well-designed techniques for weakly supervised semantic segmentation, salient region detection, and object location prediction. Detailed experiments demonstrate the effectiveness of our two-phase learning in each task.