This paper presents JavisGPT, the first unified multimodal large language model (MLLM) for joint audio-video (JAV) comprehension and generation. JavisGPT has a concise encoder-LLM-decoder architecture, which has a SyncFusion module for spatio-temporal audio-video fusion and synchrony-aware learnable queries to bridge a pretrained JAV-DiT generator. This design enables temporally coherent video-audio understanding and generation from multimodal instructions. We design an effective three-stage training pipeline consisting of multimodal pretraining, audio-video fine-tuning, and large-scale instruction-tuning, to progressively build multimodal comprehension and generation from existing vision-language models. For instruction tuning, we construct JavisInst-Omni, a high-quality instruction dataset with over 200K GPT-4o-curated audio-video-text dialogues that cover diverse and multi-level comprehension and generation scenarios. On JAV comprehension and generation benchmarks, our experiments show that JavisGPT outperforms existing MLLMs, particularly in complex and temporally synchronized settings.

Enlarging input images is a straightforward and effective approach to promote small object detection. However, simple image enlargement is significantly expensive on both computations and GPU memory. In fact, small objects are usually sparsely distributed and locally clustered. Therefore, massive feature extraction computations are wasted on the non-target background area of images. Recent works have tried to pick out target-containing regions using an extra network and perform conventional object detection, but the newly introduced computation limits their final performance. In this paper, we propose to reuse the detector's backbone to conduct feature-level object-seeking and patch-slicing, which can avoid redundant feature extraction and reduce the computation cost. Incorporating with a sparse detection head, we are able to detect small objects on high-resolution inputs (e.g., 1080P or larger) for superior performance. The resulting Efficient Small Object Detection (ESOD) approach is a generic framework, which can be applied to both CNN- and ViT-based detectors to save the computation and GPU memory costs. Extensive experiments demonstrate the efficacy and efficiency of our method. In particular, our method consistently surpasses the SOTA detectors by a large margin (e.g., 8% gains on AP) on the representative VisDrone, UAVDT, and TinyPerson datasets.

Monocular 3D detection (M3D) aims for precise 3D object localization from a single-view image which usually involves labor-intensive annotation of 3D detection boxes. Weakly supervised M3D has recently been studied to obviate the 3D annotation process by leveraging many existing 2D annotations, but it often requires extra training data such as LiDAR point clouds or multi-view images which greatly degrades its applicability and usability in various applications. We propose SKD-WM3D, a weakly supervised monocular 3D detection framework that exploits depth information to achieve M3D with a single-view image exclusively without any 3D annotations or other training data. One key design in SKD-WM3D is a self-knowledge distillation framework, which transforms image features into 3D-like representations by fusing depth information and effectively mitigates the inherent depth ambiguity in monocular scenarios with little computational overhead in inference. In addition, we design an uncertainty-aware distillation loss and a gradient-targeted transfer modulation strategy which facilitate knowledge acquisition and knowledge transfer, respectively. Extensive experiments show that SKD-WM3D surpasses the state-of-the-art clearly and is even on par with many fully supervised methods.

Inspired by the outstanding zero-shot capability of vision language models (VLMs) in image classification tasks, open-vocabulary object detection has attracted increasing interest by distilling the broad VLM knowledge into detector training. However, most existing open-vocabulary detectors learn by aligning region embeddings with categorical labels (e.g., bicycle) only, disregarding the capability of VLMs on aligning visual embeddings with fine-grained text descriptions of object parts (e.g., pedals and bells). This paper presents DVDet, a Descriptor-Enhanced Open Vocabulary Detector that introduces conditional context prompts and hierarchical textual descriptors that enable precise region-text alignment as well as open-vocabulary detection training in general. Specifically, the conditional context prompt transforms regional embeddings into image-like representations that can be directly integrated into general open vocabulary detection training. In addition, we introduce large language models as an interactive and implicit knowledge repository which enables iterative mining and refining visually oriented textual descriptors for precise region-text alignment. Extensive experiments over multiple large-scale benchmarks show that DVDet outperforms the state-of-the-art consistently by large margins.

Monocular 3D object detection aims for precise 3D localization and identification of objects from a single-view image. Despite its recent progress, it often struggles while handling pervasive object occlusions that tend to complicate and degrade the prediction of object dimensions, depths, and orientations. We design MonoMAE, a monocular 3D detector inspired by Masked Autoencoders that addresses the object occlusion issue by masking and reconstructing objects in the feature space. MonoMAE consists of two novel designs. The first is depth-aware masking that selectively masks certain parts of non-occluded object queries in the feature space for simulating occluded object queries for network training. It masks non-occluded object queries by balancing the masked and preserved query portions adaptively according to the depth information. The second is lightweight query completion that works with the depth-aware masking to learn to reconstruct and complete the masked object queries. With the proposed feature-space occlusion and completion, MonoMAE learns enriched 3D representations that achieve superior monocular 3D detection performance qualitatively and quantitatively for both occluded and non-occluded objects. Additionally, MonoMAE learns generalizable representations that can work well in new domains.

Detecting and rejecting unknown out-of-distribution (OOD) samples is critical for deployed neural networks to void unreliable predictions. In real-world scenarios, however, the efficacy of existing OOD detection methods is often impeded by the inherent imbalance of in-distribution (ID) data, which causes significant performance decline. Through statistical observations, we have identified two common challenges faced by different OOD detectors: misidentifying tail class ID samples as OOD, while erroneously predicting OOD samples as head class from ID. To explain this phenomenon, we introduce a generalized statistical framework, termed ImOOD, to formulate the OOD detection problem on imbalanced data distribution. Consequently, the theoretical analysis reveals that there exists a class-aware bias item between balanced and imbalanced OOD detection, which contributes to the performance gap. Building upon this finding, we present a unified training-time regularization technique to mitigate the bias and boost imbalanced OOD detectors across architecture designs. Our theoretically grounded method translates into consistent improvements on the representative CIFAR10-LT, CIFAR100-LT, and ImageNet-LT benchmarks against several state-of-the-art OOD detection ap- proaches. Code is available at https://github.com/alibaba/imood.

Most visual recognition studies rely heavily on crowd-labelled data in deep neural networks (DNNs) training, and they usually train a DNN for each single visual recognition task, leading to a laborious and time-consuming visual recognition paradigm. To address the two challenges, Vision-Language Models (VLMs) have been intensively investigated recently, which learns rich vision-language correlation from web-scale image-text pairs that are almost infinitely available on the Internet and enables zero-shot predictions on various visual recognition tasks with a single VLM. This paper provides a systematic review of visual language models for various visual recognition tasks, including: (1) the background that introduces the development of visual recognition paradigms; (2) the foundations of VLM that summarize the widely-adopted network architectures, pre-training objectives, and downstream tasks; (3) the widely-adopted datasets in VLM pre-training and evaluations; (4) the review and categorization of existing VLM pre-training methods, VLM transfer learning methods, and VLM knowledge distillation methods; (5) the benchmarking, analysis and discussion of the reviewed methods; (6) several research challenges and potential research directions that could be pursued in the future VLM studies for visual recognition.

Domain generalization (DG) aims to learn domain-generalizable models from one or multiple source domains that can perform well in unseen target domains. Despite its recent progress, most existing work suffers from the misalignment between the difficulty level of training samples and the capability of contemporarily trained models, leading to over-fitting or under-fitting in the trained generalization model. We design MoDify, a Momentum Difficulty framework that tackles the misalignment by balancing the seesaw between the model’s capability and the samples’ difficulties along the training process. MoDify consists of two novel designs that collaborate to fight against the misalignment while learning domain-generalizable models. The first is MoDify-based Data Augmentation which exploits an RGB Shuffle technique to generate difficulty-aware training samples on the fly. The second is MoDify-based Network Optimization which dynamically schedules the training samples for balanced and smooth learning with appropriate difficulty. Without bells and whistles, a simple implementation of MoDify achieves superior performance across multiple benchmarks. In addition, MoDify can complement existing methods as a plug-in, and it is generic and can work for different visual recognition tasks.

Without manually annotated identities, unsupervised multi-object trackers are inferior to learning reliable feature embeddings. It causes the similarity-based inter-frame association stage also be error-prone, where an uncertainty problem arises. The frame-by-frame accumulated uncertainty prevents trackers from learning the consistent feature embedding against time variation. To avoid this uncertainty problem, recent self-supervised techniques are adopted, whereas they failed to capture temporal relations. The inter-frame uncertainty still exists. In fact, this paper argues that though the uncertainty problem is inevitable, it is possible to leverage the uncertainty itself to improve the learned consistency in turn. Specifically, an uncertainty-based metric is developed to verify and rectify the risky associations. The resulting accurate pseudo-tracklets boost learning the feature consistency. And accurate tracklets can incorporate temporal information into spatial transformation. This paper proposes a tracklet-guided augmentation strategy to simulate the tracklet’s motion, which adopts a hierarchical uncertainty-based sampling mechanism for hard sample mining. The ultimate unsupervised MOT framework, namely U2MOT, is proven effective on MOT-Challenges and VisDrone-MOT benchmark. U2MOT achieves a SOTA performance among the published supervised and unsupervised trackers.

The key to OOD detection has two aspects: generalized feature representation and precise category description. Recently, vision-language models such as CLIP provide significant advances in both two issues, but constructing precise category descriptions is still in its infancy due to the absence of unseen categories. This work introduces two hierarchical contexts, namely perceptual context and spurious context, to carefully describe the precise category boundary through automatic prompt tuning. Specifically, perceptual contexts perceive the inter-category difference (e.g., cats vs apples) for current classification tasks, while spurious contexts further identify spurious (similar but exactly not) OOD samples for every single category (e.g., cats vs panthers, apples vs peaches). The two contexts hierarchically construct the precise description for a certain category, which is, first roughly classifying a sample to the predicted category and then delicately identifying whether it is truly an ID sample or actually OOD. Moreover, the precise descriptions for those categories within the vision-language framework present a novel application: CATegory-EXtensible OOD detection (CATEX). One can efficiently extend the set of recognizable categories by simply merging the hierarchical contexts learned under different sub-task settings. And extensive experiments are conducted to demonstrate CATEX’s effectiveness, robustness, and category-extensibility. For instance, CATEX consistently surpasses the rivals by a large margin with several protocols on the challenging ImageNet-1K dataset. In addition, we offer new insights on how to efficiently scale up the prompt engineering in vision-language models to recognize thousands of object categories, as well as how to incorporate large language models (like GPT-3) to boost zero-shot applications.

Temporal action proposal generation (TAPG) is a challenging task that aims to locate action instances in untrimmed videos with temporal boundaries. To evaluate the confidence of proposals, the existing works typically predict action score of proposals that are supervised by the temporal Intersection-over-Union (tIoU) between proposal and the ground-truth. In this paper, we innovatively propose a general auxiliary Background Constraint idea to further suppress low-quality proposals, by utilizing the background prediction score to restrict the confidence of proposals. In this way, the Background Constraint concept can be easily plug-and-played into existing TAPG methods (BMN, GTAD). From this perspective, we propose the Background Constraint Network (BCNet) to further take advantage of the rich information of action and background. Specifically, we introduce an Action-Background Interaction module for reliable confidence evaluation, which models the inconsistency between action and background by attention mechanisms at the frame and clip levels. Extensive experiments are conducted on two popular benchmarks, ActivityNet-1.3 and THUMOS14. The results demonstrate that our method outperforms state-of-the-art methods. Equipped with the existing action classifier, our method also achieves remarkable performance on the temporal action localization task.

Online hashing for streaming data has attracted increasing attention recently. However, most existing algorithms focus on batch inputs and instance-balanced optimization, which is limited in the single datum input case and does not match the dynamic training in online hashing. Furthermore, constantly updating the online model with new-coming samples will inevitably lead to the catastrophic forgetting problem. In this paper, we propose a novel online hashing method to handle the above-mentioned issues jointly, termed Asynchronus Teacher-Guided Bit-wise Hard Mining for Online Hashing. Firstly, to meet the needs of datum-wise online hashing, we design a novel binary codebook that is discriminative to separate different classes. Secondly, we propose a novel semantic loss (termed bit-wise attention loss) to dynamically focus on hard samples of each bit during training. Last but not least, we design a asynchronous knowledge distillation scheme to alleviate the catastrophic forgetting problem, where the teacher model is delaying updated to maintain the old knowledge, guiding the student model learning. Extensive experiments conducted on two public benchmarks demonstrate the favorable performance of our method over the state-of-the-arts.

In recent years, supervised hashing has been validated to greatly boost the performance of image retrieval. However, the label-hungry property requires massive label collection, making it intractable in practical scenarios. To liberate the model training procedure from laborious manual annotations, some unsupervised methods are proposed. However, the following two factors make unsupervised algorithms inferior to their supervised counterparts: (1) Without manually-defined labels, it is difficult to capture the semantic information across data, which is of crucial importance to guide robust binary code learning. (2) The widely adopted relaxation on binary constraints results in quantization error accumulation in the optimization procedure. To address the above-mentioned problems, in this paper, we propose a novel Unsupervised Discrete Hashing method (UDH). Specifically, to capture the semantic information, we propose a balanced graph-based semantic loss which explores the affinity priors in the original feature space. Then, we propose a novel self-supervised loss, termed orthogonal consistent loss, which can leverage semantic loss of instance and impose independence of codes. Moreover, by integrating the discrete optimization into the proposed unsupervised framework, the binary constraints are consistently preserved, alleviating the influence of quantization errors. Extensive experiments demonstrate that UDH outperforms state-of-the-art unsupervised methods for image retrieval.

Minimizing the discrepancy of feature distributions between different domains is one of the most promising directions in unsupervised domain adaptation. From the perspective of moment matching, most existing discrepancy-based methods are designed to match the second-order or lower moments, which however, have limited expression of statistical characteristic for non-Gaussian distributions. In this work, we propose a Higher-order Moment Matching (HoMM) method, and further extend the HoMM into reproducing kernel Hilbert spaces (RKHS). In particular, our proposed HoMM can perform arbitrary-order moment matching, we show that the first-order HoMM is equivalent to Maximum Mean Discrepancy (MMD) and the second-order HoMM is equivalent to Correlation Alignment (CORAL). Moreover, HoMM (order≥ 3) is expected to perform fine-grained domain alignment as higher-order statistics can approximate more complex, non-Gaussian distributions. Besides, we also exploit the pseudo-labeled target samples to learn discriminative representations in the target domain, which further improves the transfer performance. Extensive experiments are conducted, showing that our proposed HoMM consistently outperforms the existing moment matching methods by a large margin. Codes are available at https://github.com/chenchao666/HoMM-Master

Deep hashing methods have been proved to be effective and efficient for large-scale Web media search. The success of these data-driven methods largely depends on collecting sufficient labeled data, which is usually a crucial limitation in practical cases. The current solutions to this issue utilize Generative Adversarial Network (GAN) to augment data in semi-supervised learning. However, existing GAN-based methods treat image generations and hashing learning as two isolated processes, leading to generation ineffectiveness. Besides, most works fail to exploit the semantic information in unlabeled data. In this paper, we propose a novel Semi-supervised Self-pace Adversarial Hashing method, named SSAH to solve the above problems in a unified framework. The SSAH method consists of an adversarial network (A-Net) and a hashing network (H-Net). To improve the quality of generative images, first, the A-Net learns hard samples with multi-scale occlusions and multi-angle rotated deformations which compete against the learning of accurate hashing codes. Second, we design a novel self-paced hard generation policy to gradually increase the hashing difficulty of generated samples. To make use of the semantic information in unlabeled ones, we propose a semi-supervised consistent loss. The experimental results show that our method can significantly improve state-of-the-art models on both the widely-used hashing datasets and fine-grained datasets.

In recent years, hashing methods have been proved to be effective and efficient for large-scale Web media search. However, the existing general hashing methods have limited discriminative power for describing fine-grained objects that share similar overall appearance but have a subtle difference. To solve this problem, we for the first time introduce the attention mechanism to the learning of fine-grained hashing codes. Specifically, we propose a novel deep hashing model, named deep saliency hashing (DSaH), which automatically mines salient regions and learns semantic-preserving hashing codes simultaneously. DSaH is a two-step end-to-end model consisting of an attention network and a hashing network. Our loss function contains three basic components, including the semantic loss, the saliency loss, and the quantization loss. As the core of DSaH, the saliency loss guides the attention network to mine discriminative regions from pairs of images.We conduct extensive experiments on both fine-grained and general retrieval datasets for performance evaluation. Experimental results on fine-grained datasets, including Oxford Flowers, Stanford Dogs, and CUB Birds demonstrate that our DSaH performs the best for the fine-grained retrieval task and beats the strongest competitor (DTQ) by approximately 10% on both Stanford Dogs and CUB Birds. DSaH is also comparable to several state-of-the-art hashing methods on CIFAR-10 and NUS-WIDE.