Vehicle Re-identification Research Articles

Learning More in Vehicle Re-Identification: Joint Local Blur Transformation and Adversarial Network Optimization

Vehicle re-identification (ReID) tasks are an important part of smart cities and are widely used in public security. It is extremely challenging because vehicles with different identities are generated from a uniform pipeline and cannot be distinguished based only on the subtle differences in their characteristics. To enhance the network’s ability to handle the diversity of samples in order to adapt to the changing external environment, we propose a novel data augmentation method to improve its performance. Our deep learning framework mainly consists of a local blur transformation and a transformation adversarial module. In particular, we first use a random selection algorithm to find a local region of interest in an image sample. Then, the parameter generator network, a lightweight convolutional neural network, is responsible for generating four weights and then as a basis to form a filter matrix for further blur transformations. Finally, an adversarial module is employed to ensure that as much noise information as possible is added to the image sample while preserving the structure of the training datasets. Furthermore, by updating the parameter generator network, the adversarial module can help produce more appropriate and harder training samples and lead to improving the framework’s performance. Extensive experiments on datasets, i.e., VeRi-776, VehicleID, and VERI-Wild, show that our method is superior to the state-of-the-art methods.

Mixed‐attention‐based regional soft partition network for vehicle reidentification

The purpose of vehicle reidentification is to distinguish the same vehicle from different vehicles using different cameras. The main challenge of this task is the significant intra-instance discrepancy of the same vehicle from different views and the subtle inter-instance differences of similar vehicles from the same views. To address this problem, researchers have attempted to align features from different views, such as using additional metadata (colour, type, key points, mask, etc.) to improve performance. Although these attempts improve the performance of vehicle reidentification, considerable efforts are required to create additional precise annotations to the data before using these methods. This results in expensive data preparation costs, rendering these methods less convenient to use. Therefore, we propose a novel deep learning network, called a mixed-attention-based regional soft partition network, to address this problem. This network does not require additional metadata annotations; it only trains the identity label as a supervision signal and uses soft partition attention to identify specific vehicle regions. Experiments showed that the performance of the proposed method was comparable to that of the state-of-the-art method with additional annotations on the VeRri-776 and VERI-Wild datasets.

Intelligent Video Surveillance Systems for Vehicle Identification Based on Multinet Architecture

Security cameras have been proven to be particularly useful in preventing and combating crime through identification tasks. Here, two areas can be mainly distinguished: person and vehicle identification. Automatic license plate readers are the most widely used tool for vehicle identification. Although these systems are very effective, they are not reliable enough in certain circumstances. For example, due to traffic jams, vehicle position or weather conditions, the sensors cannot capture an image of the entire license plate. However, there is still a lot of additional information in the image which may also be of interest, and that needs to be analysed quickly and accurately. The correct use of the processing mechanisms can significantly reduce analysis time, increasing the efficiency of video cameras significantly. To solve this problem, we have designed a solution based on two technologies: license plate recognition and vehicle re-identification. For its development and testing, we have also created several datasets recreating a real environment. In addition, during this article, it is also possible to read about some of the main artificial intelligence techniques for these technologies, as they have served as the starting point for this research.

Cross-camera tracking of vehicle loads based on deep metric learning and edge computing

To date, vehicle load identification with a single camera has been relatively well researched. However, due to the limited view field of the camera and various assumptions in existing studies, the acquired data do not faithfully reflect the loading history of vehicle loads on long-span bridges, thus causing deviations in statistical modelling and mechanical calculations. In order to obtain more reliable data of vehicle loads, it is necessary to explore more developed methods based on existing studies. Therefore, a novel cross-camera tracking method for vehicle loads is proposed by using deep metric learning and edge computing. Vehicle load identification with a single camera is first implemented, which includes vehicle detection, 3D bounding box reconstruction, and vehicle tracking. Next, a deep metric learning network for vehicle re-identification is built and supervised learning is performed. Based on the vehicle detection, the vehicle in video frames is cropped and fed into a convolutional neural network to obtain the embedding vector. A long connection is established between neighboring edge computing nodes according to the transmission control protocol, and then cross-camera re-identification and tracking of vehicles are achieved through data communication and vehicle status vectors. Finally, the loading history of vehicle loads is sent via a wireless network to a cloud server to be stored and further utilized. After analyzing and calculating videos from an operating bridge, the results show that the average accuracy of vehicle cross-camera identification reaches 96%. Due to the ability to obtain the full loading history of vehicle loads, this method has great potential in practice.

DFR-ST: Discriminative feature representation with spatio-temporal cues for vehicle re-identification

Vehicle re-identification (re-ID) aims to discover and match the target vehicles from a gallery image set taken by different cameras on a wide range of road networks. It is crucial for lots of applications such as security surveillance and traffic management. The remarkably similar appearances of distinct vehicles and the significant changes in viewpoints and illumination conditions pose grand challenges to vehicle re-ID. Conventional solutions focus on designing global visual appearances without sufficient consideration of vehicles’ spatio-temporal relationships in different images. This paper proposes a discriminative feature representation with spatio-temporal clues (DFR-ST) for vehicle re-ID. It is capable of building robust features in the embedding space by involving appearance and spatio-temporal information. The proposed DFR-ST constructs an appearance model for a multi-grained visual representation by a two-stream architecture and a spatio-temporal metric to provide complementary information based on this multi-modal information. Experimental results on four public datasets demonstrate DFR-ST outperforms the state-of-the-art methods, which validates the effectiveness of the proposed method.

Joint Multiple Fine-grained feature for Vehicle Re-Identification

The process of recognizing the same vehicle in different scenes is called vehicle re-identification. However, due to the different locations of the surveillance cameras, there may be obstacles in the captured vehicle pictures and multiple viewpoints may make the same vehicle look different. In order to effectively reduce the interference of obstacle occlusion, multiple viewpoints, and other factors on vehicle re-identification, in this paper, we propose a multi-fine-grained feature extraction network. While retaining the global information of vehicles, we extract the finegrained features of vehicles precisely by segmenting the vehicle feature map. In addition, we introduce a new evaluation metric mean Inverse Negative Penalty (mINP) to evaluate the vehicle re-identification model more comprehensively. Our method achieves superior accuracy over the state-of-the-art methods on the challenging vehicle datasets: VeRi-776, VehicleID, and VRIC.

Exploring Spatial Significance via Hybrid Pyramidal Graph Network for Vehicle Re-Identification

Existing vehicle re-identification methods commonly use spatial pooling operations to aggregate feature maps extracted via off-the-shelf backbone networks, such as visual geometry group network (VGGNet), Google network (GoogLeNet) and residual network (ResNet). They ignore exploring the spatial significance of feature maps, eventually degrading the vehicle re-identification performance. In this paper, firstly, an innovative spatial graph network (SGN) is proposed to elaborately explore the spatial significance of feature maps. The SGN stacks multiple spatial graphs (SGs). Each SG assigns feature map’s elements as nodes and utilizes spatial neighborhood relationships to determine edges among nodes. During the SGN’s propagation, each node and its spatial neighbors on an SG are aggregated to the next SG. On the next SG, each aggregated node is re-weighted with a learnable parameter to find the significance at the corresponding location. Secondly, a novel pyramidal graph network (PGN) is designed to comprehensively explore the spatial significance of feature maps at multiple scales. The PGN organizes multiple SGNs in a pyramidal manner and makes each SGN handles feature maps of a specific scale. Finally, a hybrid pyramidal graph network (HPGN) is developed by embedding the PGN behind a ResNet-50 based backbone network. Extensive experiments on three large scale vehicle databases (i.e., VeRi776, VehicleID, and VeRi-Wild) demonstrate that the proposed HPGN is superior to state-of-the-art vehicle re-identification approaches in terms of accuracy, parameter cost, and computation cost. In addition, experiments show that the proposed PGN is universal to various backbone networks.

SJDL-Vehicle: Semi-supervised Joint Defogging Learning for Foggy Vehicle Re-identification

Vehicle re-identification (ReID) has attracted considerable attention in computer vision. Although several methods have been proposed to achieve state-of-the-art performance on this topic, re-identifying vehicle in foggy scenes remains a great challenge due to the degradation of visibility. To our knowledge, this problem is still not well-addressed so far. In this paper, to address this problem, we propose a novel training framework called Semi-supervised Joint Defogging Learning (SJDL) framework. First, the fog removal branch and the re-identification branch are integrated to perform simultaneous training. With the collaborative training scheme, defogged features generated by the defogging branch from input images can be shared to learn better representation for the re-identification branch. However, since the fog-free image of real-world data is intractable, this architecture can only be trained on the synthetic data, which may cause the domain gap problem between real-world and synthetic scenarios. To solve this problem, we design a semi-supervised defogging training scheme that can train two kinds of data alternatively in each iteration. Due to the lack of a dataset specialized for vehicle ReID in the foggy weather, we construct a dataset called FVRID which consists of real-world and synthetic foggy images to train and evaluate the performance. Experimental results show that the proposed method is effective and outperforms other existing vehicle ReID methods in the foggy weather. The code and dataset are available in https://github.com/Cihsaing/SJDL-Foggy-Vehicle-Re-Identification--AAAI2022.

Traffic Awareness Through Multiple Mobile Sensor Fusion

An intelligent traffic management system (ITMS) benefits from an accurate and global awareness of the traffic status obtained through traffic data. A growing body of literature recognizes the importance of modern vehicles (MVs) equipped with sensing devices to estimate the traffic data to generate knowledge about the traffic scene. One critical issue is how to use a low-cost sensor mounted on an MV with the purpose of increasing generalizability related to the used sensor and provide the required traffic data. In addition, the sensing coverage of a single vehicle is limited to a specific range. Multiple sensor fusion can be used to overcome these issues. This paper focuses on analyzing target vehicle’s geolocation estimated by multiple vehicles equipped with a low-cost monocular camera and proposes a new methodology for fusing those data by considering sensor estimation uncertainties. Our objective is to use the sensor fusion technique to improve the target vehicle’s geolocalization estimation accuracy and provide a more comprehensive picture of the traffic status than what can be obtained by using data from only one vehicle. Our proposed methodology includes two primary steps: (1) target vehicle re-identification to determine whether observing vehicles are identifying the same target vehicle, and (2) multiple sensor fusion to dynamically integrate the estimated geolocation of the target vehicle. Our proposed methodology provides one of the first investigations into using data from multiple moving vehicles to estimate and fuse real traffic data. The experiments show that our proposed methodology performs well to enhance geolocation estimation accuracy.

An Efficient Multiresolution Network for Vehicle Reidentification

In general, vehicle images have varying resolutions due to vehicles’ movements and different camera settings. However, most existing vehicle re-identification models are single-resolution deep networks trained with pre-uniformly resizing vehicle images, which underestimates adverse effects of varying resolutions and leads to unsatisfactory performance. A straightforward solution for dealing with varying resolutions is to train multiple vehicle re-identification models. Each model is independently trained with images of a specific resolution. However, this straightforward solution requires significant overhead and ignores intrinsic associations among different resolution images. For that, an efficient multi-resolution network (EMRN) is proposed for vehicle re-identification in this paper. Firstly, EMRN embeds a newly-designed multi-resolution feature dimension uniform module (MR-FDUM) behind a traditional backbone network (i.e., ResNet-50). As a result, the whole model can extract fixed dimensional features from different resolution images so that it can be trained with one loss function of fixed dimensional parameters rather than training multiple models. Secondly, a multi-resolution image randomly feeding strategy is designed to train EMRN, making each mini-batch data of a random resolution during the training process. Consequently, EMRN can implicitly learn collaborative multi-resolution features via only a unitary deep network. Experiments on three large scale datasets, i.e., VeRi776, VehicleID, and VRIC, demonstrate that EMRN is superior to state-of-the-art vehicle re-identification methods.

Super-resolution-based part collaboration network for vehicle re-identification

Super-resolution-based part collaboration network for vehicle re-identification

Attentive Part-Based Alignment Network for Vehicle Re-Identification

Vehicle Re-identification (Re-ID) has become a research hotspot along with the rapid development of video surveillance. Attention mechanisms are utilized in vehicle Re-ID networks but often miss the attention alignment across views. In this paper, we propose a novel Attentive Part-based Alignment Network (APANet) to learn robust, diverse, and discriminative features for vehicle Re-ID. To be specific, in order to enhance the discrimination of part features, two part-level alignment mechanisms are proposed in APANet, consisting of Part-level Orthogonality Loss (POL) and Part-level Attention Alignment Loss (PAAL). Furthermore, POL aims to maximize the diversity of part features via an orthogonal penalty among parts whilst PAAL learns view-invariant features by means of realizing attention alignment in a part-level fashion. Moreover, we propose a Multi-receptive-field Attention (MA) module to adopt an efficient and cost-effective pyramid structure. The pyramid structure is capable of employing more fine-grained and heterogeneous-scale spatial attention information through multi-receptive-field streams. In addition, the improved TriHard loss and Inter-group Feature Centroid Loss (IFCL) function are utilized to optimize both the inter-group and intra-group distance. Extensive experiments demonstrate the superiority of our model over multiple existing state-of-the-art approaches on two popular vehicle Re-ID benchmarks.

Vehicle re-identification based on keypoint segmentation of original image

Vehicle re-identification based on keypoint segmentation of original image

Vehicle Re-Identification in Aerial Imagery Based on Normalized Virtual Softmax Loss

With the development and popularization of unmanned aerial vehicles (UAVs) and surveillance cameras, vehicle re-identification (ReID) task plays an important role in the field of urban safety. The biggest challenge in the field of vehicle ReID is how to robustly learn the common visual representation of vehicle from different viewpoints while discriminate different vehicles with similar visual appearance. In order to solve this problem, this paper designs the normalized virtual softmax loss to enlarge the inter-class distance and decrease the intra-class distance, and a vehicle ReID model is proposed by jointly training the network with the proposed loss and triplet loss. In addition, we contribute a novel UAV vehicle ReID dataset from multiple viewpoint images to verify the robustness of methods. The experimental results show that comparing with the other softmax-based losses, our method achieves better performance and gets 76.70% and 98.95% in Rank-1 on VRAI and VRAI_AIR dataset, respectively.

Vehicle Re-Identification System Based on Appearance Features

Aiming at the low recognition accuracy caused by the problems of angle, illumination, and occlusion in vehicle re-identification based on deep learning, a vehicle re-identification method based on multibranch network feature extraction and two-stage retrieval feature is proposed. The multibranch feature extraction module uses ResNet-50 as the backbone network to extract the vehicle’s attribute features and apparent features, respectively, and uses the attribute features for rough retrieval. On this basis, the attribute features and apparent features are fused for fine retrieval. Through experiments, the accuracy of vehicle re-identification on Veri-776 data set and VehicleID datasets is significantly improved. In addition, based on the improved algorithm, this paper designs and develops a vehicle re-identification system, which realizes the functions of inputting file directory, selecting target image, and querying result image, and provides a visual technical scheme for vehicle re-identification and retrieval in the real scene.

Vehicle Re-Identification with Spatio-Temporal Model Leveraging by Pose View Embedding

Vehicle re-identification (Re-ID) research has intensified as numerous advancements have been made along with the rapid development of person Re-ID. In this paper, we tackle the vehicle Re-ID problem in open scenarios. This research differs from the early-stage studies that focused on a certain view, and it faces more challenges due to view variations, illumination changes, occlusions, etc. Inspired by the research of person Re-ID, we propose leveraging pose view to enhance the discrimination performance of visual features and utilizing keypoints to improve the accuracy of pose recognition. However, the visual appearance information is still limited by the changing surroundings and extremely similar appearances of vehicles. To the best of our knowledge, few methods have been aware of the spatio-temporal information to supplement visual appearance information, but they neglect the influence of the driving direction. Considering the peculiar characteristic of vehicle movements, we observe that vehicles’ poses on camera views indicating their directions are closely related to spatio-temporal cues. Consequently, we design a two-branch framework for vehicle Re-ID, including a Keypoint-based Pose Embedding Visual (KPEV) model and a Keypoint-based Pose-Guided Spatio-Temporal (KPGST) model. These models are integrated into the framework, and the results of KPEV and KPGST are fused based on a Bayesian network. Extensive experiments performed on the VeRi-776 and VehicleID datasets related to functional urban surveillance scenarios demonstrate the competitive performance of our proposed approach.

Cross-view vehicle re-identification based on graph matching

Cross-view vehicle re-identification based on graph matching

Progressive learning with multi-scale attention network for cross-domain vehicle re-identification

Progressive learning with multi-scale attention network for cross-domain vehicle re-identification

Discriminative-region attention and orthogonal-view generation model for vehicle re-identification

Vehicle re-identification (Re-ID) is urgently demanded to alleviate the pressure caused by the increasingly onerous task of urban traffic management. Multiple challenges hamper the applications of vision-based vehicle Re-ID methods: (1) The appearances of different vehicles of the same brand/model are often similar; However, (2) the appearances of the same vehicle differ significantly from different viewpoints. Previous methods mainly use manually annotated multi-attribute datasets to assist the network in getting detailed cues and in inferencing multi-view to improve the vehicle Re-ID performance. However, finely labeled vehicle datasets are usually unattainable in real application scenarios. Hence, we propose a Discriminative-Region Attention and Orthogonal-View Generation (DRA-OVG) model, which only requires identity (ID) labels to conquer the multiple challenges of vehicle Re-ID. The proposed DRA model can automatically extract the discriminative region features, which can distinguish similar vehicles. And the OVG model can generate multi-view features based on the input view features to reduce the impact of viewpoint mismatches. Finally, the distance between vehicle appearances is presented by the discriminative region features and multi-view features together. Therefore, the significance of pairwise distance measure between vehicles is enhanced in a complete feature space. Extensive experiments substantiate the effectiveness of each proposed ingredient, and experimental results indicate that our approach achieves remarkable improvements over the state-of-the-art vehicle Re-ID methods on VehicleID and VeRi-776 datasets.

Machine-Learning-Based Real-Time Multi-Camera Vehicle Tracking and Travel-Time Estimation.

Travel-time estimation of traffic flow is an important problem with critical implications for traffic congestion analysis. We developed techniques for using intersection videos to identify vehicle trajectories across multiple cameras and analyze corridor travel time. Our approach consists of (1) multi-object single-camera tracking, (2) vehicle re-identification among different cameras, (3) multi-object multi-camera tracking, and (4) travel-time estimation. We evaluated the proposed framework on real intersections in Florida with pan and fisheye cameras. The experimental results demonstrate the viability and effectiveness of our method.