Publications
Publications are listed in reverse chronological order.
* denotes equal contribution.
2025
- AAAIDriving in the Occupancy World: Vision-Centric 4D Occupancy Forecasting and Planning via World Models for Autonomous DrivingYu Yang*, Jianbiao Mei*, Yukai Ma, and 5 more authorsAAAI Conference on Artificial Intelligence (AAAI), 2025
World models envision potential future states based on various ego actions. They embed extensive knowledge about the driving environment, facilitating safe and scalable autonomous driving. Most existing methods primarily focus on either data generation or the pretraining paradigms of world models. Unlike the aforementioned prior works, we propose Drive-OccWorld, which adapts a vision-centric 4D forecasting world model to end-to-end planning for autonomous driving. Specifically, we first introduce a semantic and motion-conditional normalization in the memory module, which accumulates semantic and dynamic information from historical BEV embeddings. These BEV features are then conveyed to the world decoder for future occupancy and flow forecasting, considering both geometry and spatiotemporal modeling. Additionally, we propose injecting flexible action conditions, such as velocity, steering angle, trajectory, and commands, into the world model to enable controllable generation and facilitate a broader range of downstream applications. Furthermore, we explore integrating the generative capabilities of the 4D world model with end-to-end planning, enabling continuous forecasting of future states and the selection of optimal trajectories using an occupancy-based cost function. Extensive experiments on the nuScenes dataset demonstrate that our method can generate plausible and controllable 4D occupancy, opening new avenues for driving world generation and end-to-end planning.
2024
- PreprintDreamForge: Motion-Aware Autoregressive Video Generation for Multi-View Driving ScenesJianbiao Mei, Xuemeng Yang, Licheng Wen, and 7 more authorsarXiv preprint arXiv:2409.04003, 2024
Recent advances in diffusion models have improved controllable streetscape generation and supported downstream perception and planning tasks. However, challenges remain in accurately modeling driving scenes and generating long videos. To alleviate these issues, we propose DreamForge, an advanced diffusion-based autoregressive video generation model tailored for 3D-controllable long-term generation. To enhance the lane and foreground generation, we introduce perspective guidance and integrate object-wise position encoding to incorporate local 3D correlation and improve foreground object modeling. We also propose motion-aware temporal attention to capture motion cues and appearance changes in videos. By leveraging motion frames and an autoregressive generation paradigm, we can autoregressively generate long videos (over 200 frames) using a 7-frame model, achieving superior quality compared to the baseline in 16-frame video evaluations. Finally, we integrate our method with the realistic simulation platform DriveArena to provide more reliable open-loop and closed-loop evaluations for vision-based driving agents.
- PreprintDrivearena: A closed-loop generative simulation platform for autonomous drivingXuemeng Yang*, Licheng Wen*, Yukai Ma*, and 8 more authorsarXiv preprint arXiv:2408.00415, 2024
This paper presented DriveArena, the first high-fidelity closed-loop simulation system designed for driving agents navigating in real scenarios. DriveArena features a flexible, modular architecture, allowing for the seamless interchange of its core components: Traffic Manager, a traffic simulator capable of generating realistic traffic flow on any worldwide street map, and World Dreamer, a high-fidelity conditional generative model with infinite autoregression. This powerful synergy empowers any driving agent capable of processing real-world images to navigate in DriveArena’s simulated environment. The agent perceives its surroundings through images generated by World Dreamer and output trajectories. These trajectories are fed into Traffic Manager, achieving realistic interactions with other vehicles and producing a new scene layout. Finally, the latest scene layout is relayed back into World Dreamer, perpetuating the simulation cycle. This iterative process fosters closed-loop exploration within a highly realistic environment, providing a valuable platform for developing and evaluating driving agents across diverse and challenging scenarios. DriveArena signifies a substantial leap forward in leveraging generative image data for the driving simulation platform, opening insights for closed-loop autonomous driving.
- NeurIPSContinuously Learning, Adapting, and Improving: A Dual-Process Approach to Autonomous DrivingJianbiao Mei*, Yukai Ma*, Xuemeng Yang, and 8 more authorsIn Advances in Neural Information Processing Systems (NeurIPS), 2024
Autonomous driving has advanced significantly due to sensors, machine learning, and artificial intelligence improvements. However, prevailing methods struggle with intricate scenarios and causal relationships, hindering adaptability and interpretability in varied environments. To address the above problems, we introduce LeapAD, a novel paradigm for autonomous driving inspired by the human cognitive process. Specifically, LeapAD emulates human attention by selecting critical objects relevant to driving decisions, simplifying environmental interpretation, and mitigating decision-making complexities. Additionally, LeapAD incorporates an innovative dual-process decision-making module, which consists of an Analytic Process (System-II) for thorough analysis and reasoning, along with a Heuristic Process (System-I) for swift and empirical processing. The Analytic Process leverages its logical reasoning to accumulate linguistic driving experience, which is then transferred to the Heuristic Process by supervised fine-tuning. Through reflection mechanisms and a growing memory bank, LeapAD continuously improves itself from past mistakes in a closed-loop environment. Closed-loop testing in CARLA shows that LeapAD outperforms all methods relying solely on camera input, requiring 1-2 orders of magnitude less labeled data. Experiments also demonstrate that as the memory bank expands, the Heuristic Process with only 1.8B parameters can inherit the knowledge from a GPT-4 powered Analytic Process and achieve continuous performance improvement
- TIPCamera-based 3d semantic scene completion with sparse guidance networkJianbiao Mei, Yu Yang, Mengmeng Wang, and 5 more authorsIEEE Transactions on Image Processing (TIP), 2024
Semantic scene completion (SSC) aims to predict the semantic occupancy of each voxel in the entire 3D scene from limited observations, which is an emerging and critical task for autonomous driving. Recently, many studies have turned to camera-based SSC solutions due to the richer visual cues and cost-effectiveness of cameras. However, existing methods usually rely on sophisticated and heavy 3D models to process the lifted 3D features directly, which are not discriminative enough for clear segmentation boundaries. In this paper, we adopt the dense-sparse-dense design and propose a one-stage camera-based SSC framework, termed SGN, to propagate semantics from the semantic-aware seed voxels to the whole scene based on spatial geometry cues. Firstly, to exploit depth-aware context and dynamically select sparse seed voxels, we redesign the sparse voxel proposal network to process points generated by depth prediction directly with the coarse-to-fine paradigm. Furthermore, by designing hybrid guidance (sparse semantic and geometry guidance) and effective voxel aggregation for spatial geometry cues, we enhance the feature separation between different categories and expedite the convergence of semantic propagation. Finally, we devise the multi-scale semantic propagation module for flexible receptive fields while reducing the computation resources. Extensive experimental results on the SemanticKITTI and SSCBench-KITTI-360 datasets demonstrate the superiority of our SGN over existing state-of-the-art methods. And even our lightweight version SGN-L achieves notable scores of 14.80% mIoU and 45.45% IoU on SeamnticKITTI validation with only 12.5 M parameters and 7.16 G training memory.
- APINLearning spatiotemporal relationships with a unified framework for video object segmentationJianbiao Mei, Mengmeng Wang, Yu Yang, and 2 more authorsApplied Intelligence (APIN), 2024
Video object segmentation (VOS) has made significant progress with matching-based methods, but most approaches still show two problems. Firstly, they apply a complicated and redundant two-extractor pipeline to use more reference frames for cues, increasing the models’ parameters and complexity. Secondly, most of these methods neglect the spatial relationships (inside each frame) and do not fully model the temporal relationships (among different frames), i.e., they need adequate modeling of spatial-temporal relationships. In this paper, to address the two problems, we propose a unified transformer-based framework for VOS, a compact and unified single-extractor pipeline with strong spatial and temporal interaction ability. Specifically, to slim the common-used two-extractor pipeline while keeping the model’s effectiveness and flexibility, we design a single dynamic feature extractor with an ingenious dynamic input adapter to encode two significant inputs, i.e., reference sets (historical frames with predicted masks) and query frame (current frame), respectively. Moreover, the relationships among different frames and inside every frame are crucial for this task. We introduce a vision transformer to exploit and model both the temporal and spatial relationships simultaneously. By the cascaded design of the proposed dynamic feature extractor, transformer-based relationship module, and target-enhanced segmentation, our model implements a unified and compact pipeline for VOS. Extensive experiments demonstrate the superiority of our model over state-of-the-art methods on both DAVIS and YouTube-VOS datasets. We also explore potential solutions, such as sequence organizers, to improve the model’s efficiency. On DAVIS17 validation, we achieve 50% faster inference speed with only a slight 0.2% (J&F) drop in segmentation quality.
- PRLLiDAR video object segmentation with dynamic kernel refinementJianbiao Mei, Yu Yang, Mengmeng Wang, and 3 more authorsPattern Recognition Letters (PRL), 2024
In this paper, we formalize memory- and tracking-based methods to perform the LiDAR-based Video Object Segmentation (VOS) task, which segments points of the specific 3D target (given in the first frame) in a LiDAR sequence. LiDAR-based VOS can directly provide target-aware geometric information for practical application scenarios like behavior analysis and anticipating danger. We first construct a LiDAR-based VOS dataset named KITTI-VOS based on SemanticKITTI, which acts as a testbed and facilitates comprehensive evaluations of algorithm performance. Next, we provide two types of baselines, i.e., memory-based and tracking-based baselines, to explore this task. Specifically, the first memory-based pipeline is built on a space–time memory network equipped with the non-local spatiotemporal attention-based memory bank. We further design a more potent variant to introduce the locality into the spatiotemporal attention module by local self-attention and cross-attention modules. For the second tracking-based baseline, we modify two representative 3D object tracking methods to adapt to LiDAR-based VOS tasks. Finally, we propose a refine module that takes mask priors and generates object-aware kernels, which could boost all the baselines’ performance. We evaluate the proposed methods on the dataset and demonstrate their effectiveness.
- PreprintDQFormer: Towards Unified LiDAR Panoptic Segmentation with Decoupled QueriesYu Yang*, Jianbiao Mei*, Liang Liu, and 6 more authorsarXiv preprint arXiv:2408.15813, 2024
LiDAR panoptic segmentation, which jointly performs instance and semantic segmentation for things and stuff classes, plays a fundamental role in LiDAR perception tasks. While most existing methods explicitly separate these two segmentation tasks and utilize different branches (i.e., semantic and instance branches), some recent methods have embraced the query-based paradigm to unify LiDAR panoptic segmentation. However, the distinct spatial distribution and inherent characteristics of objects(things) and their surroundings(stuff) in 3D scenes lead to challenges, including the mutual competition of things/stuff and the ambiguity of classification/segmentation. In this paper, we propose decoupling things/stuff queries according to their intrinsic properties for individual decoding and disentangling classification/segmentation to mitigate ambiguity. To this end, we propose a novel framework dubbed DQFormer to implement semantic and instance segmentation in a unified workflow. Specifically, we design a decoupled query generator to propose informative queries with semantics by localizing things/stuff positions and fusing multi-level BEV embeddings. Moreover, a query-oriented mask decoder is introduced to decode corresponding segmentation masks by performing masked cross-attention between queries and mask embeddings. Finally, the decoded masks are combined with the semantics of the queries to produce panoptic results. Extensive experiments on nuScenes and SemanticKITTI datasets demonstrate the superiority of our DQFormer framework.
- RALLiCROcc: Teach Radar for Accurate Semantic Occupancy Prediction using LiDAR and CameraYukai Ma*, Jianbiao Mei*, Xuemeng Yang, and 6 more authorsIEEE Robotics and Automation Letters (RAL), 2024
Semantic Scene Completion (SSC) is pivotal in autonomous driving perception, frequently confronted with the complexities of weather and illumination changes. The long-term strategy involves fusing multi-modal information to bolster the system’s robustness. Radar, increasingly utilized for 3D target detection, is gradually replacing LiDAR in autonomous driving applications, offering a robust sensing alternative. In this paper, we focus on the potential of 3D radar in semantic scene completion, pioneering cross-modal refinement techniques for improved robustness against weather and illumination changes, and enhancing SSC this http URL model architecture, we propose a three-stage tight fusion approach on BEV to realize a fusion framework for point clouds and images. Based on this foundation, we designed three cross-modal distillation modules-CMRD, BRD, and PDD. Our approach enhances the performance in both radar-only (R-LiCROcc) and radar-camera (RC-LiCROcc) settings by distilling to them the rich semantic and structural information of the fused features of LiDAR and camera. Finally, our LC-Fusion (teacher model), R-LiCROcc and RC-LiCROcc achieve the best performance on the nuScenes-Occupancy dataset, with mIOU exceeding the baseline by 22.9%, 44.1%, and 15.5%, respectively.
- AAAIA Multimodal, Multi-Task Adapting Framework for Video Action RecognitionMengmeng Wang, Jiazheng Xing, Boyuan Jiang, and 6 more authorsIn Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2024
Recently, the rise of large-scale vision-language pretrained models like CLIP, coupled with the technology of Parameter-Efficient FineTuning (PEFT), has captured substantial attraction in video action recognition. Nevertheless, prevailing approaches tend to prioritize strong supervised performance at the expense of compromising the models’ generalization capabilities during transfer. In this paper, we introduce a novel Multimodal, Multi-task CLIP adapting framework named \name to address these challenges, preserving both high supervised performance and robust transferability. Firstly, to enhance the individual modality architectures, we introduce multimodal adapters to both the visual and text branches. Specifically, we design a novel visual TED-Adapter, that performs global Temporal Enhancement and local temporal Difference modeling to improve the temporal representation capabilities of the visual encoder. Moreover, we adopt text encoder adapters to strengthen the learning of semantic label information. Secondly, we design a multi-task decoder with a rich set of supervisory signals to adeptly satisfy the need for strong supervised performance and generalization within a multimodal framework. Experimental results validate the efficacy of our approach, demonstrating exceptional performance in supervised learning while maintaining strong generalization in zero-shot scenarios.
- ICRAA Coarse-to-Fine Place Recognition Approach using Attention-guided Descriptors and Overlap EstimationChencan Fu, Lin Li, Jianbiao Mei, and 4 more authorsIn 2024 IEEE International Conference on Robotics and Automation (ICRA), 2024
Place recognition is a challenging but crucial task in robotics. Current description-based methods may be limited by representation capabilities, while pairwise similarity-based methods require exhaustive searches, which is time-consuming. In this paper, we present a novel coarse-to-fine approach to address these problems, which combines BEV (Bird’s Eye View) feature extraction, coarse-grained matching and fine-grained verification. In the coarse stage, our approach utilizes an attention-guided network to generate attention-guided descriptors. We then employ a fast affinity-based candidate selection process to identify the Top-K most similar candidates. In the fine stage, we estimate pairwise overlap among the narrowed-down place candidates to determine the final match. Experimental results on the KITTI and KITTI-360 datasets demonstrate that our approach outperforms state-of-the-art methods.
- 3DVExploit Spatiotemporal Contextual Information for 3D Single Object Tracking via Memory NetworksJongwon Ra, MengMeng Wang, Jianbiao Mei, and 3 more authorsIn 2024 International Conference on 3D Vision (3DV), 2024
The point cloud-based 3D single object tracking plays an indispensable role in autonomous driving. However, the application of 3D object tracking in the real world is still challenging due to the inherent sparsity and self-occlusion of point cloud data. Therefore, it is necessary to exploit as much useful information from limited data as we can. Since 3D object tracking is a video-level task, the appearance of objects changes gradually over time, and there is rich spatiotemporal contextual information among historical frames. However, existing methods do not fully utilize this information. To address this, we propose a new method called SCTrack, which utilizes a memory-based paradigm to exploit spatiotemporal contextual information. SCTrack incorporates both long-term and short-term memory banks to store the spatiotemporal features of targets from historical frames. By doing so, the tracker can benefit from the entire video sequence and make more informed predictions. Additionally, SCTrack extracts the mask prior to augmenting the target representation, improving the target-background discriminability. Extensive experiments on KITTI, nuScenes, and Waymo Open datasets verify the effectiveness of our proposed method.
2023
- ACM MMCenterlps: Segment instances by centers for lidar panoptic segmentationJianbiao Mei*, Yu Yang*, Mengmeng Wang, and 5 more authorsIn Proceedings of the 31st ACM International Conference on Multimedia (ACM MM), 2023
This paper focuses on LiDAR Panoptic Segmentation (LPS), which has attracted more attention recently due to its broad application prospect for autonomous driving and robotics. The mainstream LPS approaches either adopt a top-down strategy relying on 3D object detectors to discover instances or utilize time-consuming heuristic clustering algorithms to group instances in a bottom-up manner. Inspired by the center representation and kernel-based segmentation, we propose a new detection-free and clustering-free framework called CenterLPS, with the center-based instance encoding and decoding paradigm. Specifically, we propose a sparse center proposal network to generate the sparse 3D instance centers, as well as center feature embedding, which can well encode characteristics of instances. Then a center-aware transformer is applied to collect the context between different center feature embedding and around centers. Moreover, we generate the kernel weights based on the enhanced center feature embedding and initialize dynamic convolutions to decode the final instance masks. Finally, a mask fusion module is devised to unify the semantic and instance predictions and improve the panoptic quality. Extensive experiments on SemanticKITTI and nuScenes demonstrate the effectiveness of our proposed center-based framework CenterLPS.
- IROSPANet: LiDAR Panoptic Segmentation with Sparse Instance Proposal and AggregationJianbiao Mei*, Yu Yang*, Mengmeng Wang, and 3 more authorsIn 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023
Reliable LiDAR panoptic segmentation (LPS), including both semantic and instance segmentation, is vital for many robotic applications, such as autonomous driving. This work proposes a new LPS framework named PANet to eliminate the dependency on the offset branch and improve the performance on large objects, which are always over-segmented by clustering algorithms. Firstly, we propose a non-learning Sparse Instance Proposal (SIP) module with the “sampling-shifting-grouping" scheme to directly group thing points into instances from the raw point cloud efficiently. More specifically, balanced point sampling is introduced to generate sparse seed points with more uniform point distribution over the distance range. And a shift module, termed bubble shifting, is proposed to shrink the seed points to the clustered centers. Then we utilize the connected component label algorithm to generate instance proposals. Furthermore, an instance aggregation module is devised to integrate potentially fragmented instances, improving the performance of the SIP module on large objects. Extensive experiments show that PANet achieves state-of-the-art performance among published works on the SemanticKITII validation and nuScenes validation for the panoptic segmentation task.
- IROSSSC-RS: Elevate LiDAR semantic scene completion with representation separation and BEV fusionJianbiao Mei, Yu Yang, Mengmeng Wang, and 3 more authorsIn 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023
Semantic scene completion (SSC) jointly predicts the semantics and geometry of the entire 3D scene, which plays an essential role in 3D scene understanding for autonomous driving systems. SSC has achieved rapid progress with the help of semantic context in segmentation. However, how to effectively exploit the relationships between the semantic context in semantic segmentation and geometric structure in scene completion remains under exploration. In this paper, we propose to solve outdoor SSC from the perspective of representation separation and BEV fusion. Specifically, we present the network, named SSC-RS, which uses separate branches with deep supervision to explicitly disentangle the learning procedure of the semantic and geometric representations. And a BEV fusion network equipped with the proposed Adaptive Representation Fusion (ARF) module is presented to aggregate the multi-scale features effectively and efficiently. Due to the low computational burden and powerful representation ability, our model has good generality while running in real-time. Extensive experiments on SemanticKITTI demonstrate our SSC-RS achieves state-of-the-art performance.
- ACM TISTFast real-time video object segmentation with a tangled memory networkJianbiao Mei*, Mengmeng Wang*, Yu Yang, and 2 more authorsACM Transactions on Intelligent Systems and Technology (ACM TIST), 2023
In this article, we present a fast real-time tangled memory network that segments the objects effectively and efficiently for semi-supervised video object segmentation (VOS). We propose a tangled reference encoder and a memory bank organization mechanism based on a state estimator to fully utilize the mask features and alleviate memory overhead and computational burden brought by the unlimited memory bank used in many memory-based methods. First, the tangled memory network exploits the mask features that uncover abundant object information like edges and contours but are not fully explored in existing methods. Specifically, a tangled two-stream reference encoder is designed to extract and fuse the features from both RGB frames and the predicted masks. Second, to indicate the quality of the predicted mask and feedback the online prediction state for organizing the memory bank, we devise a target state estimator to learn the IoU score between the predicted mask and ground truth. Moreover, to accelerate the forward process and avoid memory overflow, we use a memory bank of fixed size to store historical features by designing a new efficient memory bank organization mechanism based on the mask state score provided by the state estimator. We conduct comprehensive experiments on the public benchmarks DAVIS and YouTube-VOS, demonstrating that our method obtains competitive results while running at high speed (66 FPS on the DAVIS16-val set).
- APINExploiting semantic-level affinities with a mask-guided network for temporal action proposal in videosYu Yang, Mengmeng Wang, Jianbiao Mei, and 1 more authorApplied Intelligence (APIN), 2023
Temporal action proposal (TAP) aims to detect the action instances’ starting and ending times in untrimmed videos, which is fundamental and critical for large-scale video analysis and human action understanding. The main challenge of the temporal action proposal lies in modeling representative temporal relations in long untrimmed videos. Existing state-of-the-art methods achieve temporal modeling by building local-level, proposal-level, or global-level temporal dependencies. Local methods lack a wider receptive field, while proposal and global methods lack the focalization of learning action frames and contain background distractions. In this paper, we propose that learning semantic-level affinities can capture more practical information. Specifically, by modeling semantic associations between frames and action units, action segments (foregrounds) can aggregate supportive cues from other co-occurring actions, and nonaction clips (backgrounds) can learn the discriminations between them and action frames. To this end, we propose a novel framework named the Mask-Guided Network (MGNet) to build semantic-level temporal associations for the TAP task. Specifically, we first propose a Foreground Mask Generation (FMG) module to adaptively generate the foreground mask, representing the locations of the action units throughout the video. Second, we design a Mask-Guided Transformer (MGT) by exploiting the foreground mask to guide the self-attention mechanism to focus on and calculate semantic affinities with the foreground frames. Finally, these two modules are jointly explored in a unified framework. MGNet models the intra-semantic similarities for foregrounds, extracting supportive action cues for boundary refinement; it also builds the inter-semantic distances for backgrounds, providing the semantic gaps to suppress false positives and distractions. Extensive experiments are conducted on two challenging datasets, ActivityNet-1.3 and THUMOS14, and the results demonstrate that our method achieves superior performance.
- RALGeo-localization with transformer-based 2D-3D match networkLaijian Li, Yukai Ma, Kai Tang, and 5 more authorsIEEE Robotics and Automation Letters (RAL), 2023
This letter presents a novel method for geographical localization by registering satellite maps with LiDAR point clouds. This method includes a Transformer-based 2D-3D matching network called D-GLSNet that directly matches the LiDAR point clouds and satellite images through end-to-end learning. Without the need for feature point detection, D-GLSNet provides accurate pixel-to-point association between the LiDAR point clouds and satellite images. And then, we can easily calculate the horizontal offset (Δx,Δy) and angular deviation Δθyaw between them, thereby achieving accurate registration. To demonstrate our network’s localization potential, we have designed a Geo-localization Node (GLN) that implements geographical localization and is plug-and-play in the SLAM system. Compared to GPS, GLN is less susceptible to external interference, such as building occlusion. In urban scenarios, our proposed D-GLSNet can output high-quality matching, enabling GLN to function stably and deliver more accurate localization results. Extensive experiments on the KITTI dataset show that our D-GLSNet method achieves a mean Relative Translation Error (RTE) of 1.43 m. Furthermore, our method outperforms state-of-the-art LiDAR-based geospatial localization methods when combined with odometry.
- TNNLSActionclip: Adapting language-image pretrained models for video action recognitionMengmeng Wang, Jiazheng Xing, Jianbiao Mei, and 2 more authorsIEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2023
The canonical approach to video action recognition dictates a neural network model to do a classic and standard 1-of-N majority vote task. They are trained to predict a fixed set of predefined categories, limiting their transferability on new datasets with unseen concepts. In this article, we provide a new perspective on action recognition by attaching importance to the semantic information of label texts rather than simply mapping them into numbers. Specifically, we model this task as a video-text matching problem within a multimodal learning framework, which strengthens the video representation with more semantic language supervision and enables our model to do zero-shot action recognition without any further labeled data or parameters’ requirements. Moreover, to handle the deficiency of label texts and make use of tremendous web data, we propose a new paradigm based on this multimodal learning framework for action recognition, which we dub “pre-train, adapt and fine-tune.” This paradigm first learns powerful representations from pre-training on a large amount of web image-text or video-text data. Then, it makes the action recognition task to act more like pre-training problems via adaptation engineering. Finally, it is fine-tuned end-to-end on target datasets to obtain strong performance. We give an instantiation of the new paradigm, ActionCLIP , which not only has superior and flexible zero-shot/few-shot transfer ability but also reaches a top performance on general action recognition task, achieving 83.8% top-1 accuracy on Kinetics-400 with a ViT-B/16 as the backbone.
- PreprintCR-SFP: Learning Consistent Representation for Soft Filter PruningJingyang Xiang, Zhuangzhi Chen, Jianbiao Mei, and 3 more authorsarXiv preprint arXiv:2312.11555, 2023
Soft filter pruning (SFP) has emerged as an effective pruning technique for allowing pruned filters to update and the opportunity for them to regrow to the network. However, this pruning strategy applies training and pruning in an alternative manner, which inevitably causes inconsistent representations between the reconstructed network (R-NN) at the training and the pruned network (P-NN) at the inference, resulting in performance degradation. In this paper, we propose to mitigate this gap by learning consistent representation for soft filter pruning, dubbed as CR-SFP. Specifically, for each training step, CR-SFP optimizes the R-NN and P-NN simultaneously with different distorted versions of the same training data, while forcing them to be consistent by minimizing their posterior distribution via the bidirectional KL-divergence loss. Meanwhile, the R-NN and P-NN share backbone parameters thus only additional classifier parameters are introduced. After training, we can export the P-NN for inference. CR-SFP is a simple yet effective training framework to improve the accuracy of P-NN without introducing any additional inference cost. It can also be combined with a variety of pruning criteria and loss functions. Extensive experiments demonstrate our CR-SFP achieves consistent improvements across various CNN architectures. Notably, on ImageNet, our CR-SFP reduces more than 41.8% FLOPs on ResNet18 with 69.2% top-1 accuracy, improving SFP by 2.1% under the same training settings.
2022
- TIPDelving deeper into mask utilization in video object segmentationMengmeng Wang*, Jianbiao Mei*, Lina Liu, and 3 more authorsIEEE Transactions on Image Processing (TIP), 2022
This paper focuses on the mask utilization of video object segmentation (VOS). The mask here mains the reference masks in the memory bank, i.e., several chosen high-quality predicted masks, which are usually used with the reference frames together. The reference masks depict the edge and contour features of the target object and indicate the boundary of the target against the background, while the reference frames contain the raw RGB information of the whole image. It is obvious that the reference masks could play a significant role in the VOS, but this is not well explored yet. To tackle this, we propose to investigate the mask advantages of both the encoder and the matcher. For the encoder, we provide a unified codebase to integrate and compare eight different mask-fused encoders. Half of them are inherited or summarized from existing methods, and the other half are devised by ourselves. We find the best configuration from our design and give valuable observations from the comparison. Then, we propose a new mask-enhanced matcher to reduce the background distraction and enhance the locality of the matching process. Combining the mask-fused encoder, mask-enhanced matcher and a standard decoder, we formulate a new architecture named MaskVOS, which sufficiently exploits the mask benefits for VOS. Qualitative and quantitative results demonstrate the effectiveness of our method. We hope our exploration could raise the attention of mask utilization in VOS.
- ECCVE-nerv: Expedite neural video representation with disentangled spatial-temporal contextZizhang Li, Mengmeng Wang, Huaijin Pi, and 3 more authorsIn European Conference on Computer Vision (ECCV), 2022
Recently, the image-wise implicit neural representation of videos, NeRV, has gained popularity for its promising results and swift speed compared to regular pixel-wise implicit representations. However, the redundant parameters within the network structure can cause a large model size when scaling up for desirable performance. The key reason of this phenomenon is the coupled formulation of NeRV, which outputs the spatial and temporal information of video frames directly from the frame index input. In this paper, we propose E-NeRV, which dramatically expedites NeRV by decomposing the image-wise implicit neural representation into separate spatial and temporal context. Under the guidance of this new formulation, our model greatly reduces the redundant model parameters, while retaining the representation ability. We experimentally find that our method can improve the performance to a large extent with fewer parameters, resulting in a more than 8x faster speed on convergence.
2021
- PreprintTransvos: Video object segmentation with transformersJianbiao Mei*, Mengmeng Wang*, Yeneng Lin, and 2 more authorsarXiv preprint arXiv:2106.00588, 2021
Recently, Space-Time Memory Network (STM) based methods have achieved state-of-the-art performance in semi-supervised video object segmentation (VOS). A crucial problem in this task is how to model the dependency both among different frames and inside every frame. However, most of these methods neglect the spatial relationships (inside each frame) and do not make full use of the temporal relationships (among different frames). In this paper, we propose a new transformer-based framework, termed TransVOS, introducing a vision transformer to fully exploit and model both the temporal and spatial relationships. Moreover, most STM-based approaches employ two separate encoders to extract features of two significant inputs, i.e., reference sets (history frames with predicted masks) and query frame (current frame), respectively, increasing the models’ parameters and complexity. To slim the popular two-encoder pipeline while keeping the effectiveness, we design a single two-path feature extractor to encode the above two inputs in a unified way. Extensive experiments demonstrate the superiority of our TransVOS over state-of-the-art methods on both DAVIS and YouTube-VOS datasets.
- PreprintMail: A unified mask-image-language trimodal network for referring image segmentationZizhang Li, Mengmeng Wang, Jianbiao Mei, and 1 more authorarXiv preprint arXiv:2111.10747, 2021
Referring image segmentation is a typical multi-modal task, which aims at generating a binary mask for referent described in given language expressions. Prior arts adopt a bimodal solution, taking images and languages as two modalities within an encoder-fusion-decoder pipeline. However, this pipeline is sub-optimal for the target task for two reasons. First, they only fuse high-level features produced by uni-modal encoders separately, which hinders sufficient cross-modal learning. Second, the uni-modal encoders are pre-trained independently, which brings inconsistency between pre-trained uni-modal tasks and the target multi-modal task. Besides, this pipeline often ignores or makes little use of intuitively beneficial instance-level features. To relieve these problems, we propose MaIL, which is a more concise encoder-decoder pipeline with a Mask-Image-Language trimodal encoder. Specifically, MaIL unifies uni-modal feature extractors and their fusion model into a deep modality interaction encoder, facilitating sufficient feature interaction across different modalities. Meanwhile, MaIL directly avoids the second limitation since no uni-modal encoders are needed anymore. Moreover, for the first time, we propose to introduce instance masks as an additional modality, which explicitly intensifies instance-level features and promotes finer segmentation results. The proposed MaIL set a new state-of-the-art on all frequently-used referring image segmentation datasets, including RefCOCO, RefCOCO+, and G-Ref, with significant gains, 3%-10% against previous best methods.