Fine-grained open-vocabulary detection (FG-OVD) challenges models to detect objects described by complex, compositional language. The prevalent strategy of encoding descriptive phrases into a single, holistic vector creates significant semantic ambiguity, hindering true compositional understanding. This ambiguity manifests as two opposing failures: attribute over-representation, where a visually dominant attribute leads to mislocalization, and attribute under-representation, where the model fails to enforce a conjunctive match across all specified attributes, leading to false positives. In this paper, we propose GUIDED++, a framework that systematically resolves both issues through a dual strategy. First, we employ architectural decoupling to separate coarse-grained localization from fine-grained discrimination, effectively mitigating attribute over-representation. Second, we introduce a novel conjunctive multi-attribute verification mechanism to explicitly combat attribute under-representation. This mechanism decomposes a complex query into its atomic attribute conditions, computes a distinct similarity score for each, and uses a minimal aggregation function to enforce a strict logical ‘AND’, ensuring an object satisfies all required attributes. Extensive experiments show that GUIDED++ significantly outperforms existing methods and establishes a new state-of-the-art on challenging FG-OVD benchmarks, demonstrating a more robust approach to compositional visual reasoning.

Large vision-language models (LVLMs) have shown remarkable capabilities in visual-language understanding. Despite their success, LVLMs still suffer from generating hallucinations in complex generation tasks, leading to inconsistencies between visual inputs and generated content. To address this issue, some approaches have introduced inference-time interventions, such as contrastive decoding, to reduce overreliance on language priors. However, these approaches overlook hallucinations stemming from position bias and spurious inter-modality correlations. In this paper, we propose a Cross-Modal Attention Calibration (CMAC) method to mitigate hallucinations in LVLMs in a training-free manner. In this method, we design an Inter-Modality Decoding (IMD) module to alleviate hallucination by a novel contrastive decoding mechanism. IMD masks the value vectors associated with significant cross-modal attention weights as distortion, which addresses both uni-modality overreliance and misleading inter-modality correlations. Additionally, a Cross-Modal Position Calibration (CMPC) module shrinks the position gap of image tokens, alleviating the position bias in cross-modal attention. Experimental results on diverse hallucination benchmarks validate the superiority of our method over existing state-of-the-art techniques in reducing hallucinations for LVLM. Our code will be available.

Semi-supervised object detection (SSOD) mitigates the annotation burden in object detection by leveraging unlabeled data, providing a scalable solution for modern perception systems. Concurrently, detection transformers (DETRs) have emerged as a popular end-to-end framework, offering advantages such as non-maximum suppression (NMS)-free inference. However, existing SSOD methods are predominantly designed for conventional detectors, leaving the exploration of DETR-based SSOD largely uncharted. This paper presents a systematic study to bridge this gap. We begin by identifying two principal obstacles in semi-supervised DETR training: (1) the inherent one-to-one assignment mechanism of DETRs is highly sensitive to noisy pseudo-labels, which impedes training efficiency; and (2) the query-based decoder architecture complicates the design of an effective consistency regularization scheme, limiting further performance gains. To address these challenges, we propose Semi-DETR++, a novel framework for efficient SSOD with DETRs. Our approach introduces a stage-wise hybrid matching strategy that enhances robustness to noisy pseudo-labels by synergistically combining one-to-many and one-to-one assignments while preserving NMS-free inference. Furthermore, based on our observation of the unique layer-wise decoding behavior in DETRs, we develop a simple yet effective re-decode query consistency training method to regularize the decoder. Extensive experiments demonstrate that Semi-DETR++ enables more efficient semi-supervised learning across various DETR architectures, outperforming existing methods by significant margins. The proposed components are also flexible and versatile, showing superior generalization by readily extending to semi-supervised segmentation tasks.

Streaming video reasoning requires models to operate in a setting where history grows without bound while meaningful evidence remains scarce. In such a landscape, relevant signal is like an oasis -- small, critical, and easily lost in a desert of redundancy. Enlarging memory only widens the desert; aggressive compression dries up the oasis. The real difficulty lies in discovering where to look, not how much to remember. We therefore introduce OASIS, a novel framework for streaming video reasoning that tackles this challenge through structured, on-demand retrieval. It organizes streaming history into hierarchical events and performs reasoning as controlled refinement -- short-context inference first, followed by semantically grounded retrieval only when uncertainty arises. As the retrieval is driven by high-level intent rather than embedding similarity, the retrieve memory is substantially more accurate and less noisy. Additionally, the mechanism is plug-and-play, training-free, and compatible with any streaming MLLM. Experiments across multiple benchmarks show that OASIS achieves strong gains in long-horizon accuracy and compositional reasoning with far less memory budget.

Large vision-language models (LVLMs) have shown remarkable capabilities in visual-language understanding. Despite their success, LVLMs still suffer from generating hallucinations in complex generation tasks, leading to inconsistencies between visual inputs and generated content. To address this issue, some approaches have introduced inference-time interventions, such as contrastive decoding, to reduce overreliance on language priors. However, these approaches overlook hallucinations stemming from position bias and spurious inter-modality correlations. In this paper, we propose a Cross-Modal Attention Calibration (CMAC) method to mitigate hallucinations in LVLMs in a training-free manner. In this method, we design an Inter-Modality Decoding (IMD) module to alleviate hallucination by a novel contrastive decoding mechanism. IMD masks the value vectors associated with significant cross-modal attention weights as distortion, which addresses both uni-modality overreliance and misleading inter-modality correlations. Additionally, a Cross-Modal Position Calibration (CMPC) module shrinks the position gap of image tokens, alleviating the position bias in cross-modal attention. Experimental results on diverse hallucination benchmarks validate the superiority of our method over existing state-of-the-art techniques in reducing hallucinations for LVLM. Our code will be available at https://github.com/lijm48/IMCCD.

Fine-grained open-vocabulary object detection (FG-OVD) aims to detect novel object categories described by attribute-rich texts. While existing open-vocabulary detectors show promise at the base-category level, they underperform in fine-grained settings due to the semantic entanglement of subjects and attributes in pretrained vision-language model (VLM) embeddings -- leading to over-representation of attributes, mislocalization, and semantic drift in embedding space. We propose GUIDED, a decomposition framework specifically designed to address the semantic entanglement between subjects and attributes in fine-grained prompts. By separating object localization and fine-grained recognition into distinct pathways, GUIDED aligns each subtask with the module best suited for its respective roles. Specifically, given a fine-grained class name, we first use a language model to extract a coarse-grained subject and its descriptive attributes. Then the detector is guided solely by the subject embedding, ensuring stable localization unaffected by irrelevant or overrepresented attributes. To selectively retain helpful attributes, we introduce an attribute embedding fusion module that incorporates attribute information into detection queries in an attention-based manner. This mitigates over-representation while preserving discriminative power. Finally, a region-level attribute discrimination module compares each detected region against full fine-grained class names using a refined vision-language model with a projection head for improved alignment. Extensive experiments on FG-OVD and 3F-OVD benchmarks show that GUIDED achieves new state-of-the-art results, demonstrating the benefits of disentangled modeling and modular optimization.

Detecting out-of-distribution (OOD) inputs is critical for safely deploying deep neural networks in the open world. Recent distance-based contrastive learning methods demonstrated their effectiveness by learning improved feature representations in the embedding space. However, those methods might lead to an incomplete and ambiguous representation of a class, thereby resulting in the loss of intra-class semantic information. In this work, we propose a novel diversified multi-prototype contrastive learning framework, which preserves the semantic knowledge within each class's embedding space by introducing multiple fine-grained prototypes for each class. This preserves intrinsic features within the in-distribution data, promoting discrimination against OOD samples. We also devise an activation constraints technique to mitigate the impact of extreme activation values on other dimensions and facilitate the computation of distance-based scores. Extensive experiments on several benchmarks show that our proposed method is effective and beneficial for OOD detection, outperforming previous state-of-the-art methods.

Offset-based representation has emerged as a promising approach for modeling semantic relations between pixels and object motion, demonstrating efficacy across various computer vision tasks. In this paper, we introduce a novel one-stage multi-tasking network tailored to extend the offset-based approach to MOTS. Our proposed framework, named OffsetNet, is designed to concurrently address amodal bounding box detection, instance segmentation, and tracking. It achieves this by formulating these three tasks within a unified pixel-offset-based representation, thereby achieving excellent efficiency and encouraging mutual collaborations. OffsetNet achieves several remarkable properties: first, the encoder is empowered by a novel Memory Enhanced Linear Self-Attention (MELSA) block to efficiently aggregate spatial-temporal features; second, all tasks are decoupled fairly using three lightweight decoders that operate in a one-shot manner; third, a novel cross-frame offsets prediction module is proposed to enhance the robustness of tracking against occlusions. With these merits, OffsetNet achieves 76.83% HOTA on KITTI MOTS benchmark, which is the best result without relying on 3D detection. Furthermore, OffsetNet achieves 74.83% HOTA at 50 FPS on the KITTI MOT benchmark, which is nearly 3.3 times faster than CenterTrack with better performance. We hope our approach will serve as a solid baseline and encourage future research in this field.

Open vocabulary object detection (OVD) aims at seeking an optimal object detector capable of recognizing objects from both base and novel categories. Recent advances leverage knowledge distillation to transfer insightful knowledge from pre-trained large-scale vision-language models to the task of object detection, significantly generalizing the powerful capabilities of the detector to identify more unknown object categories. However, these methods face significant challenges in background interpretation and model overfitting and thus often result in the loss of crucial background knowledge, giving rise to sub-optimal inference performance of the detector. To mitigate these issues, we present a novel OVD framework termed LBP to propose learning background prompts to harness explored implicit background knowledge, thus enhancing the detection performance w.r.t. base and novel categories. Specifically, we devise three modules: Background Category-specific Prompt, Background Object Discovery, and Inference Probability Rectification, to empower the detector to discover, represent, and leverage implicit object knowledge explored from background proposals. Evaluation on two benchmark datasets, OV-COCO and OV-LVIS, demonstrates the superiority of our proposed method over existing state-of-the-art approaches in handling the OVD tasks.

We delve into pseudo-labeling for semi-supervised monocular 3D object detection (SSM3OD) and discover two primary issues: a misalignment between the prediction quality of 3D and 2D attributes and the tendency of depth supervision derived from pseudo-labels to be noisy, leading to significant optimization conflicts with other reliable forms of supervision. We introduce a novel decoupled pseudo-labeling (DPL) approach for SSM3OD. Our approach features a Decoupled Pseudo-label Generation (DPG) module, designed to efficiently generate pseudo-labels by separately processing 2D and 3D attributes. This module incorporates a unique homography-based method for identifying dependable pseudo-labels in BEV space, specifically for 3D attributes. Additionally, we present a DepthGradient Projection (DGP) module to mitigate optimization conflicts caused by noisy depth supervision of pseudo-labels, effectively decoupling the depth gradient and removing conflicting gradients. This dual decoupling strategy-at both the pseudo-label generation and gradient levels-significantly improves the utilization of pseudo-labels in SSM3OD. Our comprehensive experiments on the KITTI benchmark demonstrate the superiority of our method over existing approaches.

Current semi-supervised object detection (SSOD) algorithms typically assume class balanced datasets (PASCAL VOC etc.) or slightly class imbalanced datasets (MS-COCO, etc). This assumption can be easily violated since real world datasets can be extremely class imbalanced in nature, thus making the performance of semi-supervised object detectors far from satisfactory. Besides, the research for this problem in SSOD is severely under-explored. To bridge this research gap, we comprehensively study the class imbalance problem for SSOD under more challenging scenarios, thus forming the first experimental setting for class imbalanced SSOD (CI-SSOD). Moreover, we propose a simple yet effective gradient-based sampling framework that tackles the class imbalance problem from the perspective of two types of confirmation biases. To tackle confirmation bias towards majority classes, the gradient-based reweighting and gradient-based thresholding modules leverage the gradients from each class to fully balance the influence of the majority and minority classes. To tackle the confirmation bias from incorrect pseudo labels of minority classes, the class-rebalancing sampling module resamples unlabeled data following the guidance of the gradient-based reweighting module. Experiments on three proposed sub-tasks, namely MS-COCO, MS-COCO to Object365 and LVIS, suggest that our method outperforms current class imbalanced object detectors by clear margins, serving as a baseline for future research in CI-SSOD. Code will be available at https://github.com/nightkeepers/CI-SSOD.

Automatically adapting image segmentation across data sites benefits to reduce the data annotation burden in medical image analysis. Due to variations in image collection procedures, there usually exists moderate domain gap between medical image datasets from different sites. Increasing the prediction certainty is beneficial for gradually reducing the category-wise domain shift. However, uncertainty minimization naturally leads to bias towards major classes since the target object usually occupies a small portion of pixels in the input image. In this paper, we propose a gradient-rebalanced uncertainty minimization scheme which is capable of eliminating the learning bias. First, the foreground pixels and background pixels are reweighted according to the total gradient amplitude of every class. Furthermore, we devise a feature-level adaptation scheme to reduce the overall domain gap between source and target datasets, based on feature norm regularization and adversarial learning. Experiments on CT pancreas segmentation and MRI prostate segmentation validate that, our method outperforms existing cross-site adaptation algorithms by around 3% on the DICE similarity coefficient.

Most text detection methods hypothesize texts are horizontal or multi-oriented and thus define quadrangles as the basic detection unit. However, text in the wild is usually perspectively distorted or curved, which can not be easily tackled by existing approaches. In this paper, we propose a deep character embedding network (CENet) which simultaneously predicts the bounding boxes of characters and their embedding vectors, thus making text detection a simple clustering task in the character embedding space. The proposed method does not require strong assumptions of forming a straight line on general text detection, which provides flexibility on arbitrarily curved or perspectively distorted text. For character detection task, a dense prediction subnetwork is designed to obtain the confidence score and bounding boxes of characters. For character embedding task, a subnet is trained with contrastive loss to project detected characters into embedding space. The two tasks share a backbone CNN from which the multi-scale feature maps are extracted. The final text regions can be easily achieved by a thresholding process on character confidence and embedding distance of character pairs. We evaluated our method on ICDAR13, ICDAR15, MSRA-TD500, and Total Text. The proposed method achieves state-of-the-art or comparable performance on all of the datasets, and shows a substantial improvement in the irregular-text datasets, i.e. Total-Text.