Zijian Kang

Zijian Kang, Yueyang Li, Shengyu Gong et al.

Emotional Recognition in Conversation (ERC) is valuable for diagnosing health conditions such as autism and depression, and for understanding the emotions of individuals who struggle to express their feelings. Current ERC methods primarily rely on semantic, audio and visual data but face significant challenges in integrating physiological signals such as Electroencephalography (EEG). This research proposes Hypergraph Multi-Modal Learning (Hyper-MML), a novel framework for identifying emotions in conversation. Hyper-MML effectively integrates EEG with audio and video information to capture complex emotional dynamics. Firstly, we introduce an Adaptive Brain Encoder with Mutual-cross Attention (ABEMA) module for processing EEG signals. This module captures emotion-relevant features across different frequency bands and adapts to subject-specific variations through hierarchical mutual-cross attention mechanisms. Secondly, we propose an Adaptive Hypergraph Fusion Module (AHFM) to actively model the higher-order relationships among multi-modal signals in ERC. Experimental results on the EAV and AFFEC datasets demonstrate that our Hyper-MML model significantly outperforms current state-of-the-art methods. The proposed Hyper-MML can serve as an effective communication tool for healthcare professionals, enabling better engagement with patients who have difficulty expressing their emotions. The official implementation codes are available at https://github.com/NZWANG/Hyper-MML.

Zijian Kang, Weiming Zeng, Yueyang Li et al.

Low-density EEG is more suitable for wearable and IoT-based brain sensing, but sparse electrode sampling often lacks sufficient spatial information to characterize cross-regional neural activity. EEG spatial super-resolution aims to recover dense-channel EEG from sparse recordings, yet remains challenging because channel missingness typically occurs at the whole-channel level, spatiotemporal dependencies over the full electrode layout are often underexplored, and the mapping from sparse to dense signals is inherently ambiguous. To address these issues, we propose TGSD, a topology-guided state-space diffusion framework for EEG spatial super-resolution. TGSD first employs a Hierarchical Spatial Prior Encoder to learn topology-aware priors over the complete electrode layout by integrating local geometric relationships with region-level contextual information. Based on these priors and sparse observations, a Conditional State-Space Diffusion Reconstructor progressively generates missing-channel signals through reverse diffusion, while alternating temporal and channel-wise state-space modeling captures long-range temporal dynamics and inter-channel dependencies in a unified framework. Experiments on the SEED and PhysioNet MM/I datasets show that TGSD consistently outperforms representative baselines under different super-resolution factors in both reconstruction fidelity and downstream classification performance. These results demonstrate the effectiveness of combining topology-aware spatial priors with conditional diffusion for enhancing practical low-density EEG sensing in wearable and IoT scenarios. The official implementation code is available at https://github.com/jtggz/TGSD.

Yiran Qin, Chaoqun Wang, Zijian Kang et al.

In this paper, we propose a novel training strategy called SupFusion, which provides an auxiliary feature level supervision for effective LiDAR-Camera fusion and significantly boosts detection performance. Our strategy involves a data enhancement method named Polar Sampling, which densifies sparse objects and trains an assistant model to generate high-quality features as the supervision. These features are then used to train the LiDAR-Camera fusion model, where the fusion feature is optimized to simulate the generated high-quality features. Furthermore, we propose a simple yet effective deep fusion module, which contiguously gains superior performance compared with previous fusion methods with SupFusion strategy. In such a manner, our proposal shares the following advantages. Firstly, SupFusion introduces auxiliary feature-level supervision which could boost LiDAR-Camera detection performance without introducing extra inference costs. Secondly, the proposed deep fusion could continuously improve the detector's abilities. Our proposed SupFusion and deep fusion module is plug-and-play, we make extensive experiments to demonstrate its effectiveness. Specifically, we gain around 2% 3D mAP improvements on KITTI benchmark based on multiple LiDAR-Camera 3D detectors.

Yueyang Li, Weiming Zeng, Wenhao Dong et al.

Single-channel electroencephalogram (EEG) is a cost-effective, comfortable, and non-invasive method for monitoring brain activity, widely adopted by researchers, consumers, and clinicians. The increasing number and proportion of articles on single-channel EEG underscore its growing potential. This paper provides a comprehensive review of single-channel EEG, focusing on development trends, devices, datasets, signal processing methods, recent applications, and future directions. Definitions of bipolar and unipolar configurations in single-channel EEG are clarified to guide future advancements. Applications mainly span sleep staging, emotion recognition, educational research, and clinical diagnosis. Ongoing advancements of single-channel EEG in AI-based EEG generation techniques suggest potential parity or superiority over multichannel EEG performance.

Fangxun Shu, Yongjie Ye, Yue Liao et al.

We introduce SAIL-RL, a reinforcement learning (RL) post-training framework that enhances the reasoning capabilities of multimodal large language models (MLLMs) by teaching them when and how to think. Existing approaches are limited by outcome-only supervision, which rewards correct answers without ensuring sound reasoning, and by uniform thinking strategies, which often lead to overthinking on simple tasks and underthinking on complex ones. SAIL-RL addresses these challenges with a dual reward system: the Thinking Reward, which evaluates reasoning quality through factual grounding, logical coherence, and answer consistency, and the Judging Reward, which adaptively determines whether deep reasoning or direct answering is appropriate. Experiments on the state-of-the-art SAIL-VL2 show that SAIL-RL improves reasoning and multimodal understanding benchmarks at both 4B and 8B scales, achieving competitive performance against commercial closed-source models such as GPT-4o, and substantially reduces hallucinations, establishing it as a principled framework for building more reliable and adaptive MLLMs. The code will be available at https://github.com/BytedanceDouyinContent/SAIL-RL.

Hongyuan Dong, Zijian Kang, Weijie Yin et al.

In this paper, we introduce SAIL-VL (ScAlable Vision Language Model TraIning via High QuaLity Data Curation), an open-source vision language model (VLM) series achieving state-of-the-art (SOTA) performance in 2B and 8B parameters. The following three key improvements contribute to SAIL-VL's leading performance: (1) Scalable high-quality visual understanding data construction: We implement a data construction pipeline to enable hundred-million-scale high-quality recaption data annotation. The resulted dataset SAIL-Caption is validated to be of the highest data quality compared with opensource datasets. (2) Scalable Pretraining with High-Quality Visual Understanding Data: We scale SAIL-VL's pretraining budget up to 655B tokens and show that even a 2B VLM benefits from scaled up training data sizes, exhibiting logarithmic data size scaling laws in benchmark performance. (3) Scalable SFT via data quantity and complexity scaling: We curate a high-quality SFT dataset collection with leading data quantity scaling effectiveness and demonstrate that training with progressively higher-complexity data surpasses baseline one-stage training by a large margin. SAIL-VL series models achieve the highest average score in 18 widely used VLM benchmarks in our evaluation, with the 2B model takes the top position over VLMs of comparable sizes on OpenCompass 2024 (https://rank.opencompass.org.cn/leaderboard-multimodal), demonstrating robust visual comprehension abilities. SAIL-VL series models are released at HuggingFace (https://huggingface.co/BytedanceDouyinContent).

Haochen Wang, Xiangtai Li, Zilong Huang et al.

Models like OpenAI-o3 pioneer visual grounded reasoning by dynamically referencing visual regions, just like human "thinking with images". However, no benchmark exists to evaluate these capabilities holistically. To bridge this gap, we propose TreeBench (Traceable Evidence Evaluation Benchmark), a diagnostic benchmark built on three principles: (1) focused visual perception of subtle targets in complex scenes, (2) traceable evidence via bounding box evaluation, and (3) second-order reasoning to test object interactions and spatial hierarchies beyond simple object localization. Prioritizing images with dense objects, we initially sample 1K high-quality images from SA-1B, and incorporate eight LMM experts to manually annotate questions, candidate options, and answers for each image. After three stages of quality control, TreeBench consists of 405 challenging visual question-answering pairs, even the most advanced models struggle with this benchmark, where none of them reach 60% accuracy, e.g., OpenAI-o3 scores only 54.87. Furthermore, we introduce TreeVGR (Traceable Evidence Enhanced Visual Grounded Reasoning), a training paradigm to supervise localization and reasoning jointly with reinforcement learning, enabling accurate localizations and explainable reasoning pathways. Initialized from Qwen2.5-VL-7B, it improves V* Bench (+16.8), MME-RealWorld (+12.6), and TreeBench (+13.4), proving traceability is key to advancing vision-grounded reasoning. The code is available at https://github.com/Haochen-Wang409/TreeVGR.

Yueyang Li, Zijian Kang, Shengyu Gong et al.

Decoding neural visual representations from electroencephalogram (EEG)-based brain activity is crucial for advancing brain-machine interfaces (BMI) and has transformative potential for neural sensory rehabilitation. While multimodal contrastive representation learning (MCRL) has shown promise in neural decoding, existing methods often overlook semantic consistency and completeness within modalities and lack effective semantic alignment across modalities. This limits their ability to capture the complex representations of visual neural responses. We propose Neural-MCRL, a novel framework that achieves multimodal alignment through semantic bridging and cross-attention mechanisms, while ensuring completeness within modalities and consistency across modalities. Our framework also features the Neural Encoder with Spectral-Temporal Adaptation (NESTA), a EEG encoder that adaptively captures spectral patterns and learns subject-specific transformations. Experimental results demonstrate significant improvements in visual decoding accuracy and model generalization compared to state-of-the-art methods, advancing the field of EEG-based neural visual representation decoding in BMI. Codes will be available at: https://github.com/NZWANG/Neural-MCRL.

Weijie Yin, Dingkang Yang, Hongyuan Dong et al.

Vision Transformers (ViTs) are essential as foundation backbones in establishing the visual comprehension capabilities of Multimodal Large Language Models (MLLMs). Although most ViTs achieve impressive performance through image-text pair-based contrastive learning or self-supervised mechanisms, they struggle to engage in connector-based co-training directly with LLMs due to potential parameter initialization conflicts and modality semantic gaps. To address the above challenges, this paper proposes SAILViT, a gradual feature learning-enhanced ViT for facilitating MLLMs to break through performance bottlenecks in complex multimodal interactions. SAILViT achieves coarse-to-fine-grained feature alignment and world knowledge infusion with gradual feature refinement, which better serves target training demands. We perform thorough empirical analyses to confirm the powerful robustness and generalizability of SAILViT across different dimensions, including parameter sizes, model architectures, training strategies, and data scales. Equipped with SAILViT, existing MLLMs show significant and consistent performance improvements on the OpenCompass benchmark across extensive downstream tasks. SAILViT series models are released at https://huggingface.co/BytedanceDouyinContent.

Weijie Yin, Yongjie Ye, Fangxun Shu et al.

We introduce SAIL-VL2, an open-suite vision-language foundation model (LVM) for comprehensive multimodal understanding and reasoning. As the successor to SAIL-VL, SAIL-VL2 achieves state-of-the-art performance at the 2B and 8B parameter scales across diverse image and video benchmarks, demonstrating strong capabilities from fine-grained perception to complex reasoning. Its effectiveness is driven by three core innovations. First, a large-scale data curation pipeline with scoring and filtering strategies enhances both quality and distribution across captioning, OCR, QA, and video data, improving training efficiency. Second, a progressive training framework begins with a powerful pre-trained vision encoder (SAIL-ViT), advances through multimodal pre-training, and culminates in a thinking-fusion SFT-RL hybrid paradigm that systematically strengthens model capabilities. Third, architectural advances extend beyond dense LLMs to efficient sparse Mixture-of-Experts (MoE) designs. With these contributions, SAIL-VL2 demonstrates competitive performance across 106 datasets and achieves state-of-the-art results on challenging reasoning benchmarks such as MMMU and MathVista. Furthermore, on the OpenCompass leaderboard, SAIL-VL2-2B ranks first among officially released open-source models under the 4B parameter scale, while serving as an efficient and extensible foundation for the open-source multimodal community.

Shengyu Gong, Yueyang Li, Zijian Kang et al.

The accurate and efficient recognition of emotional states in oneself and others is critical, as impairments in this ability can lead to significant psychosocial difficulties. While electroencephalography (EEG) offers a powerful tool for emotion detection, current EEG-based emotion recognition (EER) methods face key limitations: insufficient model stability, limited accuracy in processing high-dimensional nonlinear EEG signals, and poor robustness against intra-subject variability and signal noise. To address these challenges, we introduce LEL (Lipschitz continuity-constrained Ensemble Learning), a novel framework that enhances EEG-based emotion recognition. By integrating Lipschitz continuity constraints, LEL ensures greater model stability and improves generalization, thereby reducing sensitivity to signal variability and noise while significantly boosting the model's overall accuracy and robustness. Its ensemble learning strategy optimizes overall performance by fusing decisions from multiple classifiers to reduce single-model bias and variance. Experimental results on three public benchmark datasets (EAV, FACED and SEED) demonstrated the LEL's state-of-the-art performance, achieving average recognition accuracies of 76.43%, 83.00% and 87.22%, respectively. The official implementation codes are released at https://github.com/NZWANG/LEL.

Zijian Kang, Peizhen Zhang, Xiangyu Zhang et al.

Knowledge distillation has shown great success in classification, however, it is still challenging for detection. In a typical image for detection, representations from different locations may have different contributions to detection targets, making the distillation hard to balance. In this paper, we propose a conditional distillation framework to distill the desired knowledge, namely knowledge that is beneficial in terms of both classification and localization for every instance. The framework introduces a learnable conditional decoding module, which retrieves information given each target instance as query. Specifically, we encode the condition information as query and use the teacher's representations as key. The attention between query and key is used to measure the contribution of different features, guided by a localization-recognition-sensitive auxiliary task. Extensive experiments demonstrate the efficacy of our method: we observe impressive improvements under various settings. Notably, we boost RetinaNet with ResNet-50 backbone from 37.4 to 40.7 mAP (+3.3) under 1x schedule, that even surpasses the teacher (40.4 mAP) with ResNet-101 backbone under 3x schedule. Code has been released on https://github.com/megvii-research/ICD.

Peizhen Zhang, Zijian Kang, Tong Yang et al.

In this paper, we propose the first self-distillation framework for general object detection, termed LGD (Label-Guided self-Distillation). Previous studies rely on a strong pretrained teacher to provide instructive knowledge that could be unavailable in real-world scenarios. Instead, we generate an instructive knowledge based only on student representations and regular labels. Our framework includes sparse label-appearance encoder, inter-object relation adapter and intra-object knowledge mapper that jointly form an implicit teacher at training phase, dynamically dependent on labels and evolving student representations. They are trained end-to-end with detector and discarded in inference. Experimentally, LGD obtains decent results on various detectors, datasets, and extensive tasks like instance segmentation. For example in MS-COCO dataset, LGD improves RetinaNet with ResNet-50 under 2x single-scale training from 36.2% to 39.0% mAP (+ 2.8%). It boosts much stronger detectors like FCOS with ResNeXt-101 DCN v2 under 2x multi-scale training from 46.1% to 47.9% (+ 1.8%). Compared with a classical teacher-based method FGFI, LGD not only performs better without requiring pretrained teacher but also reduces 51% training cost beyond inherent student learning. Codes are available at https://github.com/megvii-research/LGD.

Jiacong Wang, Zijian Kang, Haochen Wang et al.

In the field of multimodal chain-of-thought (CoT) reasoning, existing approaches predominantly rely on reasoning on pure language space, which inherently suffers from language bias and is largely confined to math or science domains. This narrow focus limits their ability to handle complex visual reasoning tasks that demand comprehensive understanding of image details. To address these limitations, this paper introduces VGR, a novel reasoning multimodal large language model (MLLM) with enhanced fine-grained visual perception capabilities. Unlike traditional MLLMs that answer the question or reasoning solely on the language space, our VGR first detects relevant regions that may help to solve problems, and then provides precise answers based on replayed image regions. To achieve this, we conduct a large-scale SFT dataset called VGR -SFT that contains reasoning data with mixed vision grounding and language deduction. The inference pipeline of VGR allows the model to choose bounding boxes for visual reference and a replay stage is introduced to integrates the corresponding regions into the reasoning process, enhancing multimodel comprehension. Experiments on the LLaVA-NeXT-7B baseline show that VGR achieves superior performance on multi-modal benchmarks requiring comprehensive image detail understanding. Compared to the baseline, VGR uses only 30\% of the image token count while delivering scores of +4.1 on MMStar, +7.1 on AI2D, and a +12.9 improvement on ChartQA.

Yueyang Li, Lei Chen, Wenhao Dong et al.

Developing interpretable models for neurodevelopmental disorders (NDDs) diagnosis presents significant challenges in effectively encoding, decoding, and integrating multimodal neuroimaging data. While many existing machine learning approaches have shown promise in brain network analysis, they typically suffer from limited interpretability, particularly in extracting meaningful biomarkers from functional magnetic resonance imaging (fMRI) data and establishing clear relationships between imaging features and demographic characteristics. Besides, current graph neural network methodologies face limitations in capturing both local and global functional connectivity patterns while simultaneously achieving theoretically principled multimodal data fusion. To address these challenges, we propose the Interpretable Information Bottleneck Heterogeneous Graph Neural Network (I2B-HGNN), a unified framework that applies information bottleneck principles to guide both brain connectivity modeling and cross-modal feature integration. This framework comprises two complementary components. The first is the Information Bottleneck Graph Transformer (IBGraphFormer), which combines transformer-based global attention mechanisms with graph neural networks through information bottleneck-guided pooling to identify sufficient biomarkers. The second is the Information Bottleneck Heterogeneous Graph Attention Network (IB-HGAN), which employs meta-path-based heterogeneous graph learning with structural consistency constraints to achieve interpretable fusion of neuroimaging and demographic data. The experimental results demonstrate that I2B-HGNN achieves superior performance in diagnosing NDDs, exhibiting both high classification accuracy and the ability to provide interpretable biomarker identification while effectively analyzing non-imaging data.

Chaoqun Wang, Yiran Qin, Zijian Kang et al.

Recent camera-based 3D object detection is limited by the precision of transforming from image to 3D feature spaces, as well as the accuracy of object localization within the 3D space. This paper aims to address such a fundamental problem of camera-based 3D object detection: How to effectively learn depth information for accurate feature lifting and object localization. Different from previous methods which directly predict depth distributions by using a supervised estimation model, we propose a cascade framework consisting of two depth-aware learning paradigms. First, a depth estimation (DE) scheme leverages relative depth information to realize the effective feature lifting from 2D to 3D spaces. Furthermore, a depth calibration (DC) scheme introduces depth reconstruction to further adjust the 3D object localization perturbation along the depth axis. In practice, the DE is explicitly realized by using both the absolute and relative depth optimization loss to promote the precision of depth prediction, while the capability of DC is implicitly embedded into the detection Transformer through a depth denoising mechanism in the training phase. The entire model training is accomplished through an end-to-end manner. We propose a baseline detector and evaluate the effectiveness of our proposal with +2.2%/+2.7% NDS/mAP improvements on NuScenes benchmark, and gain a comparable performance with 55.9%/45.7% NDS/mAP. Furthermore, we conduct extensive experiments to demonstrate its generality based on various detectors with about +2% NDS improvements.