Jiabao Wang

Yue Zhou, Xue Yang, Gefan Zhang et al.

We present an open-source toolbox, named MMRotate, which provides a coherent algorithm framework of training, inferring, and evaluation for the popular rotated object detection algorithm based on deep learning. MMRotate implements 18 state-of-the-art algorithms and supports the three most frequently used angle definition methods. To facilitate future research and industrial applications of rotated object detection-related problems, we also provide a large number of trained models and detailed benchmarks to give insights into the performance of rotated object detection. MMRotate is publicly released at https://github.com/open-mmlab/mmrotate.

Jiabao Wang, Yuming Chen, Zhaohui Zheng et al.

Knowledge Distillation (KD) has been validated as an effective model compression technique for learning compact object detectors. Existing state-of-the-art KD methods for object detection are mostly based on feature imitation. In this paper, we present a general and effective prediction mimicking distillation scheme, called CrossKD, which delivers the intermediate features of the student's detection head to the teacher's detection head. The resulting cross-head predictions are then forced to mimic the teacher's predictions. This manner relieves the student's head from receiving contradictory supervision signals from the annotations and the teacher's predictions, greatly improving the student's detection performance. Moreover, as mimicking the teacher's predictions is the target of KD, CrossKD offers more task-oriented information in contrast with feature imitation. On MS COCO, with only prediction mimicking losses applied, our CrossKD boosts the average precision of GFL ResNet-50 with 1x training schedule from 40.2 to 43.7, outperforming all existing KD methods. In addition, our method also works well when distilling detectors with heterogeneous backbones. Code is available at https://github.com/jbwang1997/CrossKD.

Yuming Chen, Xinbin Yuan, Jiabao Wang et al.

We aim at providing the object detection community with an efficient and performant object detector, termed YOLO-MS. The core design is based on a series of investigations on how multi-branch features of the basic block and convolutions with different kernel sizes affect the detection performance of objects at different scales. The outcome is a new strategy that can significantly enhance multi-scale feature representations of real-time object detectors. To verify the effectiveness of our work, we train our YOLO-MS on the MS COCO dataset from scratch without relying on any other large-scale datasets, like ImageNet or pre-trained weights. Without bells and whistles, our YOLO-MS outperforms the recent state-of-the-art real-time object detectors, including YOLO-v7, RTMDet, and YOLO-v8. Taking the XS version of YOLO-MS as an example, it can achieve an AP score of 42+% on MS COCO, which is about 2% higher than RTMDet with the same model size. Furthermore, our work can also serve as a plug-and-play module for other YOLO models. Typically, our method significantly advances the APs, APl, and AP of YOLOv8-N from 18%+, 52%+, and 37%+ to 20%+, 55%+, and 40%+, respectively, with even fewer parameters and MACs. Code and trained models are publicly available at https://github.com/FishAndWasabi/YOLO-MS. We also provide the Jittor version at https://github.com/NK-JittorCV/nk-yolo.

Rui Zhang, Ziqi Wang, Yang Li et al.

Few-shot learning (FSL) is one of the significant and hard problems in the field of image classification. However, in contrast to the rapid development of the visible light dataset, the progress in SAR target image classification is much slower. The lack of unified benchmark is a key reason for this phenomenon, which may be severely overlooked by the current literature. The researchers of SAR target image classification always report their new results on their own datasets and experimental setup. It leads to inefficiency in result comparison and impedes the further progress of this area. Motivated by this observation, we propose a novel few-shot SAR image classification benchmark (FewSAR) to address this issue. FewSAR consists of an open-source Python code library of 15 classic methods in three categories for few-shot SAR image classification. It provides an accessible and customizable testbed for different few-shot SAR image classification task. To further understanding the performance of different few-shot methods, we establish evaluation protocols and conduct extensive experiments within the benchmark. By analyzing the quantitative results and runtime under the same setting, we observe that the accuracy of metric learning methods can achieve the best results. Meta-learning methods and fine-tuning methods perform poorly on few-shot SAR images, which is primarily due to the bias of existing datasets. We believe that FewSAR will open up a new avenue for future research and development, on real-world challenges at the intersection of SAR image classification and few-shot deep learning. We will provide our code for the proposed FewSAR at https://github.com/solarlee/FewSAR.

Jiabao Wang, Yang Li, Xiu-Shen Wei et al.

Unsupervised learning technology has caught up with or even surpassed supervised learning technology in general object classification (GOC) and person re-identification (re-ID). However, it is found that the unsupervised learning of fine-grained visual classification (FGVC) is more challenging than GOC and person re-ID. In order to bridge the gap between unsupervised and supervised learning for FGVC, we investigate the essential factors (including feature extraction, clustering, and contrastive learning) for the performance gap between supervised and unsupervised FGVC. Furthermore, we propose a simple, effective, and practical method, termed as UFCL, to alleviate the gap. Three key issues are concerned and improved: First, we introduce a robust and powerful backbone, ResNet50-IBN, which has an ability of domain adaptation when we transfer ImageNet pre-trained models to FGVC tasks. Next, we propose to introduce HDBSCAN instead of DBSCAN to do clustering, which can generate better clusters for adjacent categories with fewer hyper-parameters. Finally, we propose a weighted feature agent and its updating mechanism to do contrastive learning by using the pseudo labels with inevitable noise, which can improve the optimization process of learning the parameters of the network. The effectiveness of our UFCL is verified on CUB-200-2011, Oxford-Flowers, Oxford-Pets, Stanford-Dogs, Stanford-Cars and FGVC-Aircraft datasets. Under the unsupervised FGVC setting, we achieve state-of-the-art results, and analyze the key factors and the important parameters to provide a practical guidance.

Jiabao Wang, Zhaojiang Liu, Qiang Meng et al.

Occupancy prediction, aiming at predicting the occupancy status within voxelized 3D environment, is quickly gaining momentum within the autonomous driving community. Mainstream occupancy prediction works first discretize the 3D environment into voxels, then perform classification on such dense grids. However, inspection on sample data reveals that the vast majority of voxels is unoccupied. Performing classification on these empty voxels demands suboptimal computation resource allocation, and reducing such empty voxels necessitates complex algorithm designs. To this end, we present a novel perspective on the occupancy prediction task: formulating it as a streamlined set prediction paradigm without the need for explicit space modeling or complex sparsification procedures. Our proposed framework, called OPUS, utilizes a transformer encoder-decoder architecture to simultaneously predict occupied locations and classes using a set of learnable queries. Firstly, we employ the Chamfer distance loss to scale the set-to-set comparison problem to unprecedented magnitudes, making training such model end-to-end a reality. Subsequently, semantic classes are adaptively assigned using nearest neighbor search based on the learned locations. In addition, OPUS incorporates a suite of non-trivial strategies to enhance model performance, including coarse-to-fine learning, consistent point sampling, and adaptive re-weighting, etc. Finally, compared with current state-of-the-art methods, our lightest model achieves superior RayIoU on the Occ3D-nuScenes dataset at near 2x FPS, while our heaviest model surpasses previous best results by 6.1 RayIoU.

Daniel Dauner, Valentin Charraut, Bastian Berle et al.

The pursuit of autonomous driving has produced one of the richest sensor data collections in all of robotics. However, its scale and diversity remain largely untapped. Each dataset adopts different 2D and 3D modalities, such as cameras, lidar, ego states, annotations, traffic lights, and HD maps, with different rates and synchronization schemes. They come in fragmented formats requiring complex dependencies that cannot natively coexist in the same development environment. Further, major inconsistencies in annotation conventions prevent training or measuring generalization across multiple datasets. We present 123D, an open-source framework that unifies such multi-modal driving data through a single API. To handle synchronization, we store each modality as an independent timestamped event stream with no prescribed rate, enabling synchronous or asynchronous access across arbitrary datasets. Using 123D, we consolidate eight real-world driving datasets spanning 3,300 hours and 90,000 kilometers, together with a synthetic dataset with configurable collection scripts, and provide tools for data analysis and visualization. We conduct a systematic study comparing annotation statistics and assessing each dataset's pose and calibration accuracy. Further, we showcase two applications 123D enables: cross-dataset 3D object detection transfer and reinforcement learning for planning, and offer recommendations for future directions. Code and documentation are available at https://github.com/kesai-labs/py123d.

Jiabao Wang, Qiang Meng, Guochao Liu et al.

In autonomous driving, the temporal stability of 3D object detection greatly impacts the driving safety. However, the detection stability cannot be accessed by existing metrics such as mAP and MOTA, and consequently is less explored by the community. To bridge this gap, this work proposes Stability Index (SI), a new metric that can comprehensively evaluate the stability of 3D detectors in terms of confidence, box localization, extent, and heading. By benchmarking state-of-the-art object detectors on the Waymo Open Dataset, SI reveals interesting properties of object stability that have not been previously discovered by other metrics. To help models improve their stability, we further introduce a general and effective training strategy, called Prediction Consistency Learning (PCL). PCL essentially encourages the prediction consistency of the same objects under different timestamps and augmentations, leading to enhanced detection stability. Furthermore, we examine the effectiveness of PCL with the widely-used CenterPoint, and achieve a remarkable SI of 86.00 for vehicle class, surpassing the baseline by 5.48. We hope our work could serve as a reliable baseline and draw the community's attention to this crucial issue in 3D object detection. Codes will be made publicly available.

Yining Shi, Kun Jiang, Qiang Meng et al. · tsinghua

World models are critical for autonomous driving to simulate environmental dynamics and generate synthetic data. Existing methods struggle to disentangle ego-vehicle motion (perspective shifts) from scene evolvement (agent interactions), leading to suboptimal predictions. Instead, we propose to separate environmental changes from ego-motion by leveraging the scene-centric coordinate systems. In this paper, we introduce COME: a framework that integrates scene-centric forecasting Control into the Occupancy world ModEl. Specifically, COME first generates ego-irrelevant, spatially consistent future features through a scene-centric prediction branch, which are then converted into scene condition using a tailored ControlNet. These condition features are subsequently injected into the occupancy world model, enabling more accurate and controllable future occupancy predictions. Experimental results on the nuScenes-Occ3D dataset show that COME achieves consistent and significant improvements over state-of-the-art (SOTA) methods across diverse configurations, including different input sources (ground-truth, camera-based, fusion-based occupancy) and prediction horizons (3s and 8s). For example, under the same settings, COME achieves 26.3% better mIoU metric than DOME and 23.7% better mIoU metric than UniScene. These results highlight the efficacy of disentangled representation learning in enhancing spatio-temporal prediction fidelity for world models. Code and videos will be available at https://github.com/synsin0/COME.

Jiabao Wang, Hongyu Zhou, Yuanbo Yang et al.

Despite rapid progress, autonomous driving algorithms remain notoriously fragile under Out-of-Distribution (OOD) conditions. We identify a critical decoupling failure in current research: the lack of distinction between appearance-based shifts, such as weather and lighting, and structural scene changes. This leaves a fundamental question unanswered: Is the planner failing because of complex road geometry, or simply because it is raining? To resolve this, we establish navdream, a high-fidelity robustness benchmark leveraging generative pixel-aligned style transfer. By creating a visual stress test with negligible geometric deviation, we isolate the impact of appearance on driving performance. Our evaluation reveals that existing planning algorithms often show significant degradation under OOD appearance conditions, even when the underlying scene structure remains consistent. To bridge this gap, we propose a universal perception interface leveraging a frozen visual foundation model (DINOv3). By extracting appearance-invariant features as a stable interface for the planner, we achieve exceptional zero-shot generalization across diverse planning paradigms, including regression-based, diffusion-based, and scoring-based models. Our plug-and-play solution maintains consistent performance across extreme appearance shifts without requiring further fine-tuning. The benchmark and code will be made available.

Yao Chen, Jiabao Wang, Peichao Wang et al.

Low-resolution fine-grained image classification has recently made significant progress, largely thanks to the super-resolution techniques and knowledge distillation methods. However, these approaches lead to an exponential increase in the number of parameters and computational complexity of models. In order to solve this problem, in this letter, we propose a Vision Mamba Distillation (ViMD) approach to enhance the effectiveness and efficiency of low-resolution fine-grained image classification. Concretely, a lightweight super-resolution vision Mamba classification network (SRVM-Net) is proposed to improve its capability for extracting visual features by redesigning the classification sub-network with Mamba modeling. Moreover, we design a novel multi-level Mamba knowledge distillation loss boosting the performance, which can transfer prior knowledge obtained from a High-resolution Vision Mamba classification Network (HRVM-Net) as a teacher into the proposed SRVM-Net as a student. Extensive experiments on seven public fine-grained classification datasets related to benchmarks confirm our ViMD achieves a new state-of-the-art performance. While having higher accuracy, ViMD outperforms similar methods with fewer parameters and FLOPs, which is more suitable for embedded device applications. Code is available at https://github.com/boa2004plaust/ViMD.

Gong Cheng, Jiabao Wang, Ke Li et al.

Oriented object detection is a practical and challenging task in remote sensing image interpretation. Nowadays, oriented detectors mostly use horizontal boxes as intermedium to derive oriented boxes from them. However, the horizontal boxes are inclined to get small Intersection-over-Unions (IoUs) with ground truths, which may have some undesirable effects, such as introducing redundant noise, mismatching with ground truths, detracting from the robustness of detectors, etc. In this paper, we propose a novel Anchor-free Oriented Proposal Generator (AOPG) that abandons horizontal box-related operations from the network architecture. AOPG first produces coarse oriented boxes by a Coarse Location Module (CLM) in an anchor-free manner and then refines them into high-quality oriented proposals. After AOPG, we apply a Fast R-CNN head to produce the final detection results. Furthermore, the shortage of large-scale datasets is also a hindrance to the development of oriented object detection. To alleviate the data insufficiency, we release a new dataset on the basis of our DIOR dataset and name it DIOR-R. Massive experiments demonstrate the effectiveness of AOPG. Particularly, without bells and whistles, we achieve the accuracy of 64.41%, 75.24% and 96.22% mAP on the DIOR-R, DOTA and HRSC2016 datasets respectively. Code and models are available at https://github.com/jbwang1997/AOPG.

Xingxing Xie, Gong Cheng, Jiabao Wang et al.

Current state-of-the-art two-stage detectors generate oriented proposals through time-consuming schemes. This diminishes the detectors' speed, thereby becoming the computational bottleneck in advanced oriented object detection systems. This work proposes an effective and simple oriented object detection framework, termed Oriented R-CNN, which is a general two-stage oriented detector with promising accuracy and efficiency. To be specific, in the first stage, we propose an oriented Region Proposal Network (oriented RPN) that directly generates high-quality oriented proposals in a nearly cost-free manner. The second stage is oriented R-CNN head for refining oriented Regions of Interest (oriented RoIs) and recognizing them. Without tricks, oriented R-CNN with ResNet50 achieves state-of-the-art detection accuracy on two commonly-used datasets for oriented object detection including DOTA (75.87% mAP) and HRSC2016 (96.50% mAP), while having a speed of 15.1 FPS with the image size of 1024$\times$1024 on a single RTX 2080Ti. We hope our work could inspire rethinking the design of oriented detectors and serve as a baseline for oriented object detection. Code is available at https://github.com/jbwang1997/OBBDetection.

Hongyu Zhou, Longzhong Lin, Jiabao Wang et al.

In the past few decades, autonomous driving algorithms have made significant progress in perception, planning, and control. However, evaluating individual components does not fully reflect the performance of entire systems, highlighting the need for more holistic assessment methods. This motivates the development of HUGSIM, a closed-loop, photo-realistic, and real-time simulator for evaluating autonomous driving algorithms. We achieve this by lifting captured 2D RGB images into the 3D space via 3D Gaussian Splatting, improving the rendering quality for closed-loop scenarios, and building the closed-loop environment. In terms of rendering, We tackle challenges of novel view synthesis in closed-loop scenarios, including viewpoint extrapolation and 360-degree vehicle rendering. Beyond novel view synthesis, HUGSIM further enables the full closed simulation loop, dynamically updating the ego and actor states and observations based on control commands. Moreover, HUGSIM offers a comprehensive benchmark across more than 70 sequences from KITTI-360, Waymo, nuScenes, and PandaSet, along with over 400 varying scenarios, providing a fair and realistic evaluation platform for existing autonomous driving algorithms. HUGSIM not only serves as an intuitive evaluation benchmark but also unlocks the potential for fine-tuning autonomous driving algorithms in a photorealistic closed-loop setting.

Yupeng Zhou, Lianghua Huang, Zhifan Wu et al.

In this work, we propose Mutual Forcing, a framework for fast autoregressive audio-video generation with long-horizon audio-video synchronization. Our approach addresses two key challenges: joint audio-video modeling and fast autoregressive generation. To ease joint audio-video optimization, we adopt a two-stage training strategy: we first train uni-modal generators and then couple them into a unified audio-video model for joint training on paired data. For streaming generation, we ask whether a native fast causal audio-video model can be trained directly, instead of following existing streaming distillation pipelines that typically train a bidirectional model first and then convert it into a causal generator through multiple distillation stages. Our answer is Mutual Forcing, which builds directly on native autoregressive model and integrates few-step and multi-step generation within a single weight-shared model, enabling self-distillation and improved training-inference consistency. The multi-step mode improves the few-step mode via self-distillation, while the few-step mode generates historical context during training to improve training-inference consistency; because the two modes share parameters, these two effects reinforce each other within a single model. Compared with prior approaches such as Self-Forcing, Mutual Forcing removes the need for an additional bidirectional teacher model, supports more flexible training sequence lengths, reduces training overhead, and allows the model to improve directly from real paired data rather than a fixed teacher. Experiments show that Mutual Forcing matches or surpasses strong baselines that require around 50 sampling steps while using only 4 to 8 steps, demonstrating substantial advantages in both efficiency and quality. The project page is available at https://mutualforcing.github.io.

Jaime Spencer, Fabio Tosi, Matteo Poggi et al.

This paper discusses the results of the third edition of the Monocular Depth Estimation Challenge (MDEC). The challenge focuses on zero-shot generalization to the challenging SYNS-Patches dataset, featuring complex scenes in natural and indoor settings. As with the previous edition, methods can use any form of supervision, i.e. supervised or self-supervised. The challenge received a total of 19 submissions outperforming the baseline on the test set: 10 among them submitted a report describing their approach, highlighting a diffused use of foundational models such as Depth Anything at the core of their method. The challenge winners drastically improved 3D F-Score performance, from 17.51% to 23.72%.

Qiang Meng, Xiao Wang, JiaBao Wang et al.

Despite its compactness and information integrity, the range view representation of LiDAR data rarely occurs as the first choice for 3D perception tasks. In this work, we further push the envelop of the range-view representation with a novel multi-task framework, achieving unprecedented 3D detection performances. Our proposed Small, Versatile, and Mighty (SVM) network utilizes a pure convolutional architecture to fully unleash the efficiency and multi-tasking potentials of the range view representation. To boost detection performances, we first propose a range-view specific Perspective Centric Label Assignment (PCLA) strategy, and a novel View Adaptive Regression (VAR) module to further refine hard-to-predict box properties. In addition, our framework seamlessly integrates semantic segmentation and panoptic segmentation tasks for the LiDAR point cloud, without extra modules. Among range-view-based methods, our model achieves new state-of-the-art detection performances on the Waymo Open Dataset. Especially, over 10 mAP improvement over convolutional counterparts can be obtained on the vehicle class. Our presented results for other tasks further reveal the multi-task capabilities of the proposed small but mighty framework.

Peichao Wang, Jiabao Wang, Yao Chen et al.

The data-driven method for infrared small target detection (IRSTD) has achieved promising results. However, due to the small scale of infrared small target datasets and the limited number of pixels occupied by the targets themselves, it is a challenging task for deep learning methods to directly learn from these samples. Utilizing human expert knowledge to assist deep learning methods in better learning is worthy of exploration. To effectively guide the model to focus on targets' spatial features, this paper proposes the Local Contrast Attention Enhanced infrared small target detection Network (LCAE-Net), combining prior knowledge with data-driven deep learning methods. LCAE-Net is a U-shaped neural network model which consists of two developed modules: a Local Contrast Enhancement (LCE) module and a Channel Attention Enhancement (CAE) module. The LCE module takes advantages of prior knowledge, leveraging handcrafted convolution operator to acquire Local Contrast Attention (LCA), which could realize background suppression while enhance the potential target region, thus guiding the neural network to pay more attention to potential infrared small targets' location information. To effectively utilize the response information throughout downsampling progresses, the CAE module is proposed to achieve the information fusion among feature maps' different channels. Experimental results indicate that our LCAE-Net outperforms existing state-of-the-art methods on the three public datasets NUDT-SIRST, NUAA-SIRST, and IRSTD-1K, and its detection speed could reach up to 70 fps. Meanwhile, our model has a parameter count and Floating-Point Operations (FLOPs) of 1.945M and 4.862G respectively, which is suitable for deployment on edge devices.

Honghui Chen, Yuhang Qiu, Jiabao Wang et al.

Internal Language Model (LM)-based methods use permutation language modeling (PLM) to solve the error correction caused by conditional independence in external LM-based methods. However, random permutations of human interference cause fit oscillations in the model training, and Iterative Refinement (IR) operation to improve multimodal information decoupling also introduces additional overhead. To address these issues, this paper proposes the Hierarchical Attention autoregressive Model with Adaptive Permutation (HAAP) to enhance the location-context-image interaction capability, improving autoregressive generalization with internal LM. First, we propose Implicit Permutation Neurons (IPN) to generate adaptive attention masks to dynamically exploit token dependencies. The adaptive masks increase the diversity of training data and prevent model dependency on a specific order. It reduces the training overhead of PLM while avoiding training fit oscillations. Second, we develop Cross-modal Hierarchical Attention mechanism (CHA) to couple context and image features. This processing establishes rich positional semantic dependencies between context and image while avoiding IR. Extensive experimental results show the proposed HAAP achieves state-of-the-art (SOTA) performance in terms of accuracy, complexity, and latency on several datasets.

Shipeng Ji, Yang Li, Ruizhi Fu et al.

As deep learning applications extensively increase by leaps and bounds, their interpretability has become increasingly prominent. As a universal property, chirality exists widely in nature, and applying it to the explanatory research of deep learning may be helpful to some extent. Inspired by a recent study that used CNN (convolutional neural network), which applied visual chirality, to distinguish whether an image is flipped or not. In this paper, we study feature chirality innovatively, which shows how the statistics of deep learning models' feature data are changed by training. We rethink the feature-level chirality property, propose the feature chirality, and give the measure. Our analysis of feature chirality on AlexNet, VGG, and ResNet reveals similar but surprising results, including the prevalence of feature chirality in these models, the initialization methods of the models do not affect feature chirality. Our work shows that feature chirality implies model evaluation, interpretability of the model, and model parameters optimization.

Tao Xu, Wang Dang, Jiabao Wang et al.

One of the most significant challenges of EEG-based emotion recognition is the cross-subject EEG variations, leading to poor performance and generalizability. This paper proposes a novel EEG-based emotion recognition model called the domain adversarial graph attention model (DAGAM). The basic idea is to generate a graph to model multichannel EEG signals using biological topology. Graph theory can topologically describe and analyze relationships and mutual dependency between channels of EEG. Then, unlike other graph convolutional networks, self-attention pooling is applied to benefit salient EEG feature extraction from the graph, which effectively improves the performance. Finally, after graph pooling, the domain adversarial based on the graph is employed to identify and handle EEG variation across subjects, efficiently reaching good generalizability. We conduct extensive evaluations on two benchmark datasets (SEED and SEED IV) and obtain state-of-the-art results in subject-independent emotion recognition. Our model boosts the SEED accuracy to 92.59% (4.69% improvement) with the lowest standard deviation of 3.21% (2.92% decrements) and SEED IV accuracy to 80.74% (6.90% improvement) with the lowest standard deviation of 4.14% (3.88% decrements) respectively.

Fang Gao, Jiabao Wang, Jun Yu et al.

Monocular depth estimation (MDE) is a fundamental task in many applications such as scene understanding and reconstruction. However, most of the existing methods rely on accurately labeled datasets. A weakly-supervised framework based on attention nested U-net (ANU) named as ANUW is introduced in this paper for cases with wrong labels. The ANUW is trained end-to-end to convert an input single RGB image into a depth image. It consists of a dense residual network structure, an adaptive weight channel attention (AWCA) module, a patch second non-local (PSNL) module and a soft label generation method. The dense residual network is the main body of the network to encode and decode the input. The AWCA module can adaptively adjust the channel weights to extract important features. The PSNL module implements the spatial attention mechanism through a second-order non-local method. The proposed soft label generation method uses the prior knowledge of the dataset to produce soft labels to replace false ones. The proposed ANUW is trained on a defective monocular depth dataset and the trained model is tested on three public datasets, and the results demonstrate the superiority of ANUW in comparison with the state-of-the-art MDE methods.

Jiabao Wang, Yang Li, Shanshan Jiao et al.

Convolutional neural networks have shown outstanding effectiveness in person re-identification (re-ID). However, the models always have large number of parameters and much computation for mobile application. In order to relieve this problem, we propose a novel grafted network (GraftedNet), which is designed by grafting a high-accuracy rootstock and a light-weighted scion. The rootstock is based on the former parts of ResNet-50 to provide a strong baseline, while the scion is a new designed module, composed of the latter parts of SqueezeNet, to compress the parameters. To extract more discriminative feature representation, a joint multi-level and part-based feature is proposed. In addition, to train GraftedNet efficiently, we propose an accompanying learning method, by adding an accompanying branch to train the model in training and removing it in testing for saving parameters and computation. On three public person re-ID benchmarks (Market1501, DukeMTMC-reID and CUHK03), the effectiveness of GraftedNet are evaluated and its components are analyzed. Experimental results show that the proposed GraftedNet achieves 93.02%, 85.3% and 76.2% in Rank-1 and 81.6%, 74.7% and 71.6% in mAP, with only 4.6M parameters.

Jiabao Wang, Yang Li, Yangshuo Zhang et al.

Convolutional neural networks with multiple branches have recently been proved highly effective in person re-identification (re-ID). Researchers design multi-branch networks using part models, yet they always attribute the effectiveness to multiple parts. In addition, existing multi-branch networks always have isomorphic branches, which lack structural diversity. In order to improve this problem, we propose a novel Heterogeneous Branch and Multi-level Classification Network (HBMCN), which is designed based on the pre-trained ResNet-50 model. A new heterogeneous branch, SE-Res-Branch, is proposed based on the SE-Res module, which consists of the Squeeze-and-Excitation block and the residual block. Furthermore, a new multi-level classification joint objective function is proposed for the supervised learning of HBMCN, whereby multi-level features are extracted from multiple high-level layers and concatenated to represent a person. Based on three public person re-ID benchmarks (Market1501, DukeMTMC-reID and CUHK03), experimental results show that the proposed HBMCN reaches 94.4%, 85.7% and 73.8% in Rank-1, and 85.7%, 74.6% and 69.0% in mAP, achieving a state-of-the-art performance. Further analysis demonstrates that the specially designed heterogeneous branch performs better than an isomorphic branch, and multi-level classification provides more discriminative features compared to single-level classification. As a result, HBMCN provides substantial further improvements in person re-ID tasks.

Jiabao Wang, Yang Li, Zhuang Miao

In person re-identification (re-ID), the key task is feature representation, which is used to compute distance or similarity in prediction. Person re-ID achieves great improvement when deep learning methods are introduced to tackle this problem. The features extracted by convolutional neural networks (CNN) are more effective and discriminative than the hand-crafted features. However, deep feature extracted by a single CNN network is not robust enough in testing stage. To improve the ability of feature representation, we propose a new ensemble network (EnsembleNet) by dividing a single network into multiple end-to-end branches. The ensemble feature is obtained by concatenating each of the branch features to represent a person. EnsembleNet is designed based on ResNet-50 and its backbone shares most of the parameters for saving computation and memory cost. Experimental results show that our EnsembleNet achieves the state-of-the-art performance on the public Market1501, DukeMTMC-reID and CUHK03 person re-ID benchmarks.