Teli Ma

Peng Gao, Teli Ma, Hongsheng Li et al.

Vision Transformers (ViT) become widely-adopted architectures for various vision tasks. Masked auto-encoding for feature pretraining and multi-scale hybrid convolution-transformer architectures can further unleash the potentials of ViT, leading to state-of-the-art performances on image classification, detection and semantic segmentation. In this paper, our ConvMAE framework demonstrates that multi-scale hybrid convolution-transformer can learn more discriminative representations via the mask auto-encoding scheme. However, directly using the original masking strategy leads to the heavy computational cost and pretraining-finetuning discrepancy. To tackle the issue, we adopt the masked convolution to prevent information leakage in the convolution blocks. A simple block-wise masking strategy is proposed to ensure computational efficiency. We also propose to more directly supervise the multi-scale features of the encoder to boost multi-scale features. Based on our pretrained ConvMAE models, ConvMAE-Base improves ImageNet-1K finetuning accuracy by 1.4% compared with MAE-Base. On object detection, ConvMAE-Base finetuned for only 25 epochs surpasses MAE-Base fined-tuned for 100 epochs by 2.9% box AP and 2.2% mask AP respectively. Code and pretrained models are available at https://github.com/Alpha-VL/ConvMAE.

Sheng Xu, Yanjing Li, Teli Ma et al.

Binary neural networks (BNNs) have received ever-increasing popularity for their great capability of reducing storage burden as well as quickening inference time. However, there is a severe performance drop compared with real-valued networks, due to its intrinsic frequent weight oscillation during training. In this paper, we introduce a Resilient Binary Neural Network (ReBNN) to mitigate the frequent oscillation for better BNNs' training. We identify that the weight oscillation mainly stems from the non-parametric scaling factor. To address this issue, we propose to parameterize the scaling factor and introduce a weighted reconstruction loss to build an adaptive training objective. For the first time, we show that the weight oscillation is controlled by the balanced parameter attached to the reconstruction loss, which provides a theoretical foundation to parameterize it in back propagation. Based on this, we learn our ReBNN by calculating the balanced parameter based on its maximum magnitude, which can effectively mitigate the weight oscillation with a resilient training process. Extensive experiments are conducted upon various network models, such as ResNet and Faster-RCNN for computer vision, as well as BERT for natural language processing. The results demonstrate the overwhelming performance of our ReBNN over prior arts. For example, our ReBNN achieves 66.9% Top-1 accuracy with ResNet-18 backbone on the ImageNet dataset, surpassing existing state-of-the-arts by a significant margin. Our code is open-sourced at https://github.com/SteveTsui/ReBNN.

Sheng Xu, Yanjing Li, Bohan Zeng et al.

Knowledge distillation (KD) has been proven to be useful for training compact object detection models. However, we observe that KD is often effective when the teacher model and student counterpart share similar proposal information. This explains why existing KD methods are less effective for 1-bit detectors, caused by a significant information discrepancy between the real-valued teacher and the 1-bit student. This paper presents an Information Discrepancy-aware strategy (IDa-Det) to distill 1-bit detectors that can effectively eliminate information discrepancies and significantly reduce the performance gap between a 1-bit detector and its real-valued counterpart. We formulate the distillation process as a bi-level optimization formulation. At the inner level, we select the representative proposals with maximum information discrepancy. We then introduce a novel entropy distillation loss to reduce the disparity based on the selected proposals. Extensive experiments demonstrate IDa-Det's superiority over state-of-the-art 1-bit detectors and KD methods on both PASCAL VOC and COCO datasets. IDa-Det achieves a 76.9% mAP for a 1-bit Faster-RCNN with ResNet-18 backbone. Our code is open-sourced on https://github.com/SteveTsui/IDa-Det.

Sheng Xu, Yanjing Li, Tiancheng Wang et al.

Binary Neural Networks (BNNs) show great promise for real-world embedded devices. As one of the critical steps to achieve a powerful BNN, the scale factor calculation plays an essential role in reducing the performance gap to their real-valued counterparts. However, existing BNNs neglect the intrinsic bilinear relationship of real-valued weights and scale factors, resulting in a sub-optimal model caused by an insufficient training process. To address this issue, Recurrent Bilinear Optimization is proposed to improve the learning process of BNNs (RBONNs) by associating the intrinsic bilinear variables in the back propagation process. Our work is the first attempt to optimize BNNs from the bilinear perspective. Specifically, we employ a recurrent optimization and Density-ReLU to sequentially backtrack the sparse real-valued weight filters, which will be sufficiently trained and reach their performance limits based on a controllable learning process. We obtain robust RBONNs, which show impressive performance over state-of-the-art BNNs on various models and datasets. Particularly, on the task of object detection, RBONNs have great generalization performance. Our code is open-sourced on https://github.com/SteveTsui/RBONN .

Teli Ma, Mengmeng Wang, Jimin Xiao et al.

Siamese network has been a de facto benchmark framework for 3D LiDAR object tracking with a shared-parametric encoder extracting features from template and search region, respectively. This paradigm relies heavily on an additional matching network to model the cross-correlation/similarity of the template and search region. In this paper, we forsake the conventional Siamese paradigm and propose a novel single-branch framework, SyncTrack, synchronizing the feature extracting and matching to avoid forwarding encoder twice for template and search region as well as introducing extra parameters of matching network. The synchronization mechanism is based on the dynamic affinity of the Transformer, and an in-depth analysis of the relevance is provided theoretically. Moreover, based on the synchronization, we introduce a novel Attentive Points-Sampling strategy into the Transformer layers (APST), replacing the random/Farthest Points Sampling (FPS) method with sampling under the supervision of attentive relations between the template and search region. It implies connecting point-wise sampling with the feature learning, beneficial to aggregating more distinctive and geometric features for tracking with sparse points. Extensive experiments on two benchmark datasets (KITTI and NuScenes) show that SyncTrack achieves state-of-the-art performance in real-time tracking.

Teli Ma, Rong Li, Junwei Liang

With the success of Large Language Models (LLMs), many Generative Vision-Language Models (GVLMs) have been constructed via multimodal instruction tuning. However, the performance of GVLMs in multimodal compositional reasoning remains under-explored. In this paper, we examine both the evaluation metrics (VisualGPTScore, etc.) and current benchmarks for evaluating the compositionality of GVLMs. We identify the syntactical bias in current benchmarks, which is exploited by the linguistic capability of GVLMs. The bias renders VisualGPTScore an insufficient metric for assessing GVLMs. To combat this, we first introduce a SyntaxBias Score, leveraging LLMs to quantify such bias for mitigation. A challenging new task is subsequently added to evaluate the robustness of GVLMs against inherent inclination toward syntactical correctness. Using the bias-mitigated datasets and the new task, we propose a novel benchmark, namely SyntActically DE-biased benchmark (SADE). Our study provides an unbiased benchmark for the compositionality of GVLMs, facilitating future research in this direction (Code and dataset are available at https://github.com/TeleeMa/SADE).

Rong Li, ShiJie Li, Xieyuanli Chen et al.

LiDAR semantic segmentation plays a crucial role in enabling autonomous driving and robots to understand their surroundings accurately and robustly. A multitude of methods exist within this domain, including point-based, range-image-based, polar-coordinate-based, and hybrid strategies. Among these, range-image-based techniques have gained widespread adoption in practical applications due to their efficiency. However, they face a significant challenge known as the ``many-to-one'' problem caused by the range image's limited horizontal and vertical angular resolution. As a result, around 20% of the 3D points can be occluded. In this paper, we present TFNet, a range-image-based LiDAR semantic segmentation method that utilizes temporal information to address this issue. Specifically, we incorporate a temporal fusion layer to extract useful information from previous scans and integrate it with the current scan. We then design a max-voting-based post-processing technique to correct false predictions, particularly those caused by the ``many-to-one'' issue. We evaluated the approach on two benchmarks and demonstrated that the plug-in post-processing technique is generic and can be applied to various networks.

Teli Ma, Jia Zheng, Zifan Wang et al.

Vision-Language-Action (VLA) models have emerged as a promising paradigm for robot learning, but their representations are still largely inherited from static image-text pretraining, leaving physical dynamics to be learned from comparatively limited action data. Generative video models, by contrast, encode rich spatiotemporal structure and implicit physics, making them a compelling foundation for robotic manipulation. But their potentials are not fully explored in the literature. To bridge the gap, we introduce DiT4DiT, an end-to-end Video-Action Model that couples a video Diffusion Transformer with an action Diffusion Transformer in a unified cascaded framework. Instead of relying on reconstructed future frames, DiT4DiT extracts intermediate denoising features from the video generation process and uses them as temporally grounded conditions for action prediction. We further propose a dual flow-matching objective with decoupled timesteps and noise scales for video prediction, hidden-state extraction, and action inference, enabling coherent joint training of both modules. Across simulation and real-world benchmarks, DiT4DiT achieves state-of-the-art results, reaching average success rates of 98.6% on LIBERO and 50.8% on RoboCasa GR1 while using substantially less training data. On the Unitree G1 robot, it also delivers superior real-world performance and strong zero-shot generalization. Importantly, DiT4DiT improves sample efficiency by over 10x and speeds up convergence by up to 7x, demonstrating that video generation can serve as an effective scaling proxy for robot policy learning. We release code and models at https://dit4dit.github.io/.

Teli Ma, Shijie Geng, Mengmeng Wang et al.

The visual world naturally exhibits a long-tailed distribution of open classes, which poses great challenges to modern visual systems. Existing approaches either perform class re-balancing strategies or directly improve network modules to address the problem. However, they still train models with a finite set of predefined labels, limiting their supervision information and restricting their transferability to novel instances. Recent advances in large-scale contrastive visual-language pretraining shed light on a new pathway for visual recognition. With open-vocabulary supervisions, pretrained contrastive vision-language models learn powerful multimodal representations that are promising to handle data deficiency and unseen concepts. By calculating the semantic similarity between visual and text inputs, visual recognition is converted to a vision-language matching problem. Inspired by this, we propose BALLAD to leverage contrastive vision-language models for long-tailed recognition. We first continue pretraining the vision-language backbone through contrastive learning on a specific long-tailed target dataset. Afterward, we freeze the backbone and further employ an additional adapter layer to enhance the representations of tail classes on balanced training samples built with re-sampling strategies. Extensive experiments have been conducted on three popular long-tailed recognition benchmarks. As a result, our simple and effective approach sets the new state-of-the-art performances and outperforms competitive baselines with a large margin. Code is released at https://github.com/gaopengcuhk/BALLAD.

Peng Gao, Shijie Geng, Renrui Zhang et al.

Large-scale contrastive vision-language pre-training has shown significant progress in visual representation learning. Unlike traditional visual systems trained by a fixed set of discrete labels, a new paradigm was introduced in \cite{radford2021learning} to directly learn to align images with raw texts in an open-vocabulary setting. On downstream tasks, a carefully chosen text prompt is employed to make zero-shot predictions.~To avoid non-trivial prompt engineering, context optimization \cite{zhou2021coop} has been proposed to learn continuous vectors as task-specific prompts with few-shot training examples.~In this paper, we show that there is an alternative path to achieve better vision-language models other than prompt tuning.~While prompt tuning is for the textual inputs, we propose CLIP-Adapter to conduct fine-tuning with feature adapters on either visual or language branch. Specifically, CLIP-Adapter adopts an additional bottleneck layer to learn new features and performs residual-style feature blending with the original pre-trained features.~As a consequence, CLIP-Adapter is able to outperform context optimization while maintains a simple design. Experiments and extensive ablation studies on various visual classification tasks demonstrate the effectiveness of our approach. Code is released at t https://github.com/gaopengcuhk/CLIP-Adapter.

Teli Ma, Zifan Wang, Jiaming Zhou et al.

Inferring affordable (i.e., graspable) parts of arbitrary objects based on human specifications is essential for robots advancing toward open-vocabulary manipulation. Current grasp planners, however, are hindered by limited vision-language comprehension and time-consuming 3D radiance modeling, restricting real-time, open-vocabulary interactions with objects. To address these limitations, we propose GLOVER, a unified Generalizable Open-Vocabulary Affordance Reasoning framework, which fine-tunes the Large Language Models (LLMs) to predict the visual affordance of graspable object parts within RGB feature space. We compile a dataset of over 10,000 images from human-object interactions, annotated with unified visual and linguistic affordance labels, to enable multi-modal fine-tuning. GLOVER inherits world knowledge and common-sense reasoning from LLMs, facilitating more fine-grained object understanding and sophisticated tool-use reasoning. To enable effective real-world deployment, we present Affordance-Aware Grasping Estimation (AGE), a non-parametric grasp planner that aligns the gripper pose with a superquadric surface derived from affordance data. In evaluations across 30 table-top real-world scenes, GLOVER achieves success rates of 86.0% in part identification and 76.3% in grasping, with speeds approximately 29 times faster in affordance reasoning and 40 times faster in grasping pose estimation than the previous state-of-the-art. We also validate the generalization across embodiments, showing effectiveness in humanoid robots with dexterous hands.

Jiaming Zhou, Ke Ye, Jiayi Liu et al.

The generalization capabilities of vision-language-action (VLA) models to unseen tasks are crucial to achieving general-purpose robotic manipulation in open-world settings. However, the cross-task generalization capabilities of existing VLA models remain significantly underexplored. To address this gap, we introduce AGNOSTOS, a novel simulation benchmark designed to rigorously evaluate cross-task zero-shot generalization in manipulation. AGNOSTOS comprises 23 unseen manipulation tasks for testing, distinct from common training task distributions, and incorporates two levels of generalization difficulty to assess robustness. Our systematic evaluation reveals that current VLA models, despite being trained on diverse datasets, struggle to generalize effectively to these unseen tasks. To overcome this limitation, we propose Cross-Task In-Context Manipulation (X-ICM), a method that conditions large language models (LLMs) on in-context demonstrations from seen tasks to predict action sequences for unseen tasks. Additionally, we introduce a dynamics-guided sample selection strategy that identifies relevant demonstrations by capturing cross-task dynamics. On AGNOSTOS, X-ICM significantly improves cross-task zero-shot generalization performance over leading VLAs. We believe AGNOSTOS and X-ICM will serve as valuable tools for advancing general-purpose robotic manipulation.

Teli Ma, Jia Zheng, Zifan Wang et al.

Learning manipulation skills from human demonstration videos offers a promising path toward generalizable and interpretable robotic intelligence-particularly through the lens of actionable affordances. However, transferring such knowledge remains challenging due to: 1) a lack of large-scale datasets with precise affordance annotations, and 2) insufficient exploration of affordances in diverse manipulation contexts. To address these gaps, we introduce HOVA-500K, a large-scale, affordance-annotated dataset comprising 500,000 images across 1,726 object categories and 675 actions. We also release a standardized benchmarking suite for multi-modal affordance reasoning. Built upon HOVA-500K, we present GLOVER++, a global-to-local affordance training framework that effectively transfers actionable affordance knowledge from human demonstrations to downstream open-vocabulary reasoning tasks. GLOVER++ achieves state-of-the-art results on the HOVA-500K benchmark and demonstrates strong generalization across diverse downstream robotic manipulation tasks. By explicitly modeling actionable affordances, GLOVER++ facilitates robust transfer across scenes, modalities, and tasks. We hope that HOVA-500K and the GLOVER++ framework will serve as valuable resources for bridging the gap between human demonstrations and robotic manipulation capabilities.

Dengyang Jiang, Zanyi Wang, Hengzhuang Li et al.

Building a generalized affordance grounding model to identify actionable regions on objects is vital for real-world applications. Existing methods to train the model can be divided into weakly and fully supervised ways. However, the former method requires a complex training framework design and can not infer new actions without an auxiliary prior. While the latter often struggle with limited annotated data and components trained from scratch despite being simpler. This study focuses on fully supervised affordance grounding and overcomes its limitations by proposing AffordanceSAM, which extends SAM's generalization capacity in segmentation to affordance grounding. Specifically, we design an affordance-adaption module and curate a coarse-to-fine annotated dataset called C2F-Aff to thoroughly transfer SAM's robust performance to affordance in a three-stage training manner. Experimental results confirm that AffordanceSAM achieves state-of-the-art (SOTA) performance on the AGD20K benchmark and exhibits strong generalized capacity.

Mengmeng Wang, Teli Ma, Shuo Xin et al.

Visual Object Tracking (VOT) is an attractive and significant research area in computer vision, which aims to recognize and track specific targets in video sequences where the target objects are arbitrary and class-agnostic. The VOT technology could be applied in various scenarios, processing data of diverse modalities such as RGB, thermal infrared and point cloud. Besides, since no one sensor could handle all the dynamic and varying environments, multi-modal VOT is also investigated. This paper presents a comprehensive survey of the recent progress of both single-modal and multi-modal VOT, especially the deep learning methods. Specifically, we first review three types of mainstream single-modal VOT, including RGB, thermal infrared and point cloud tracking. In particular, we conclude four widely-used single-modal frameworks, abstracting their schemas and categorizing the existing inheritors. Then we summarize four kinds of multi-modal VOT, including RGB-Depth, RGB-Thermal, RGB-LiDAR and RGB-Language. Moreover, the comparison results in plenty of VOT benchmarks of the discussed modalities are presented. Finally, we provide recommendations and insightful observations, inspiring the future development of this fast-growing literature.

Jiaming Zhou, Teli Ma, Kun-Yu Lin et al.

Learning generalizable visual representations across different embodied environments is essential for effective robotic manipulation in real-world scenarios. However, the limited scale and diversity of robot demonstration data pose a significant challenge. Recent research has explored leveraging large-scale human activity data for pre-training, but the substantial morphological differences between humans and robots introduce a significant human-robot domain discrepancy, hindering the generalization of these models to downstream manipulation tasks. To overcome this, we propose a novel adaptation paradigm that leverages readily available paired human-robot video data to bridge the domain gap. Our method employs a human-robot contrastive alignment loss to align the semantics of human and robot videos, adapting pre-trained models to the robot domain in a parameter-efficient manner. Experiments on 20 simulated tasks across two different benchmarks and five real-world tasks demonstrate significant improvements. These results span both single-task and language-conditioned multi-task settings, evaluated using two different pre-trained models. Compared to existing pre-trained models, our adaptation method improves the average success rate by over 7% across multiple tasks on both simulated benchmarks and real-world evaluations.

Teli Ma, Jiaming Zhou, Zifan Wang et al.

Developing robots capable of executing various manipulation tasks, guided by natural language instructions and visual observations of intricate real-world environments, remains a significant challenge in robotics. Such robot agents need to understand linguistic commands and distinguish between the requirements of different tasks. In this work, we present Sigma-Agent, an end-to-end imitation learning agent for multi-task robotic manipulation. Sigma-Agent incorporates contrastive Imitation Learning (contrastive IL) modules to strengthen vision-language and current-future representations. An effective and efficient multi-view querying Transformer (MVQ-Former) for aggregating representative semantic information is introduced. Sigma-Agent shows substantial improvement over state-of-the-art methods under diverse settings in 18 RLBench tasks, surpassing RVT by an average of 5.2% and 5.9% in 10 and 100 demonstration training, respectively. Sigma-Agent also achieves 62% success rate with a single policy in 5 real-world manipulation tasks. The code will be released upon acceptance.

Mengmeng Wang, Teli Ma, Xingxing Zuo et al.

3D LiDAR-based single object tracking (SOT) has gained increasing attention as it plays a crucial role in 3D applications such as autonomous driving. The central problem is how to learn a target-aware representation from the sparse and incomplete point clouds. In this paper, we propose a novel Correlation Pyramid Network (CorpNet) with a unified encoder and a motion-factorized decoder. Specifically, the encoder introduces multi-level self attentions and cross attentions in its main branch to enrich the template and search region features and realize their fusion and interaction, respectively. Additionally, considering the sparsity characteristics of the point clouds, we design a lateral correlation pyramid structure for the encoder to keep as many points as possible by integrating hierarchical correlated features. The output features of the search region from the encoder can be directly fed into the decoder for predicting target locations without any extra matcher. Moreover, in the decoder of CorpNet, we design a motion-factorized head to explicitly learn the different movement patterns of the up axis and the x-y plane together. Extensive experiments on two commonly-used datasets show our CorpNet achieves state-of-the-art results while running in real-time.

Sheng Xu, Yanjing Li, Teli Ma et al.

Vision transformers (ViTs) have demonstrated great potential in various visual tasks, but suffer from expensive computational and memory cost problems when deployed on resource-constrained devices. In this paper, we introduce a ternary vision transformer (TerViT) to ternarize the weights in ViTs, which are challenged by the large loss surface gap between real-valued and ternary parameters. To address the issue, we introduce a progressive training scheme by first training 8-bit transformers and then TerViT, and achieve a better optimization than conventional methods. Furthermore, we introduce channel-wise ternarization, by partitioning each matrix to different channels, each of which is with an unique distribution and ternarization interval. We apply our methods to popular DeiT and Swin backbones, and extensive results show that we can achieve competitive performance. For example, TerViT can quantize Swin-S to 13.1MB model size while achieving above 79% Top-1 accuracy on ImageNet dataset.

Teli Ma, Mingyuan Mao, Honghui Zheng et al.

Object detection with Transformers (DETR) has achieved a competitive performance over traditional detectors, such as Faster R-CNN. However, the potential of DETR remains largely unexplored for the more challenging task of arbitrary-oriented object detection problem. We provide the first attempt and implement Oriented Object DEtection with TRansformer ($\bf O^2DETR$) based on an end-to-end network. The contributions of $\rm O^2DETR$ include: 1) we provide a new insight into oriented object detection, by applying Transformer to directly and efficiently localize objects without a tedious process of rotated anchors as in conventional detectors; 2) we design a simple but highly efficient encoder for Transformer by replacing the attention mechanism with depthwise separable convolution, which can significantly reduce the memory and computational cost of using multi-scale features in the original Transformer; 3) our $\rm O^2DETR$ can be another new benchmark in the field of oriented object detection, which achieves up to 3.85 mAP improvement over Faster R-CNN and RetinaNet. We simply fine-tune the head mounted on $\rm O^2DETR$ in a cascaded architecture and achieve a competitive performance over SOTA in the DOTA dataset.

Mingyuan Mao, Renrui Zhang, Honghui Zheng et al.

Transformers with remarkable global representation capacities achieve competitive results for visual tasks, but fail to consider high-level local pattern information in input images. In this paper, we present a generic Dual-stream Network (DS-Net) to fully explore the representation capacity of local and global pattern features for image classification. Our DS-Net can simultaneously calculate fine-grained and integrated features and efficiently fuse them. Specifically, we propose an Intra-scale Propagation module to process two different resolutions in each block and an Inter-Scale Alignment module to perform information interaction across features at dual scales. Besides, we also design a Dual-stream FPN (DS-FPN) to further enhance contextual information for downstream dense predictions. Without bells and whistles, the proposed DS-Net outperforms DeiT-Small by 2.4% in terms of top-1 accuracy on ImageNet-1k and achieves state-of-the-art performance over other Vision Transformers and ResNets. For object detection and instance segmentation, DS-Net-Small respectively outperforms ResNet-50 by 6.4% and 5.5% in terms of mAP on MSCOCO 2017, and surpasses the previous state-of-the-art scheme, which significantly demonstrates its potential to be a general backbone in vision tasks. The code will be released soon.

Wenyu Zhao, Teli Ma, Xuan Gong et al.

Edge computing is promising to become one of the next hottest topics in artificial intelligence because it benefits various evolving domains such as real-time unmanned aerial systems, industrial applications, and the demand for privacy protection. This paper reviews recent advances on binary neural network (BNN) and 1-bit CNN technologies that are well suitable for front-end, edge-based computing. We introduce and summarize existing work and classify them based on gradient approximation, quantization, architecture, loss functions, optimization method, and binary neural architecture search. We also introduce applications in the areas of computer vision and speech recognition and discuss future applications for edge computing.