Neural Network For Person Re-identification Research Articles

Choice of Activation Function in Convolutional Neural Networks for Person Re-Identification in Video Surveillance Systems

Choice of Activation Function in Convolutional Neural Networks for Person Re-Identification in Video Surveillance Systems

Multi-scale feature pyramid and multi-branch neural network for person re-identification

Multi-scale feature pyramid and multi-branch neural network for person re-identification

Learning global and local features using graph neural networks for person re-identification

Person re-identification (re-id) is the task of recognizing an individual across non-overlapping camera views. Some approaches only rely on extracting global appearance features from images and fail to consider people’s local body information (head, foot, body shape), which can be used as complementary. Other techniques combine local and global features but rely on external information such as pedestrian attributes or human pose to locate and align the local regions. This strategy increases the learning difficulty and is not efficient or robust to real-world scenarios. In this paper, we propose an end-to-end deep learning framework to overcome these limitations. Our method combines the global and local feature representations of a pedestrian and captures the body structural information by modeling the spatial relation of patches using graph neural networks. We also represent the relationships between probe–gallery pairs using a graph neural network and propose incorporating a scoring function to mine a correspondence for local regions. Experimental results on several datasets validate the effectiveness of the proposed method.

List-wise learning-to-rank with convolutional neural networks for person re-identification

In this paper, we present a novel machine learning-based image ranking approach using Convolutional Neural Networks (CNN). Our proposed method relies on a similarity metric learning algorithm operating on lists of image examples and a loss function taking into account the ranking in these lists with respect to different query images. This comprises two major contributions: (1) Rank lists instead of image pairs or triplets are used for training, thus integrating more explicitly the order of similarity and relations between sets of images. (2) A weighting is introduced in the loss function based on two evaluation measures: the mean average precision and the rank 1 score. We evaluated our approach on two different computer vision applications that are commonly formulated as ranking problems: person re-identification and image retrieval with several public benchmarks and showed that our new loss function outperforms other common functions and that our method achieves state-of-the-art performance compared to existing approaches from the literature.

Deep-Facial Feature-Based Person Re-identification for Authentication in Surveillance Applications

Nowadays, a large network of cameras is predominantly used in public places which provide enormous video data. These data are monitored manually and may be utilized only when the need arises to ascertain the facts. Automating the system can improve the quality of surveillance and be useful for high-level surveillance tasks like person identification, suspicious activity detection or undesirable event prediction for timely alerts. In this paper, we proposed a model that can Re-identify a person from a single camera tracking environment. This system will automatically extract face features of the person and generate the Unique Id for each person when it enters for the first time in the monitored area. Its face features are stored in the database which will help to Re-identify the person whenever the same person appears again. The challenges faced by the system are occlusion, pose, light conditions, and face orientation. The proposed system highlights, effect of different deep neural networks for Person Re-identification and compares based on the accuracy, GPU usage, Speed, Number of faces detected by overcoming the challenges like illumination and occlusion. The advantage of the system is it doesn′t require the database of people in advance for recognition and it will be helpful for criminal identification for crime control and prevention.

Asymmetric filtering-based dense convolutional neural network for person re-identification combined with Joint Bayesian and re-ranking

Person re-identification aims at matching individuals across multiple camera views under surveillance systems. The major challenges lie in the lack of spatial and temporal cues, which makes it difficult to cope with large variations of lighting conditions, viewing angles, body poses and occlusions. How to extract multimodal features including facial features, physical features, behavioral features, color features, etc is still a fundamental problem in person re-identification. In this paper, we propose a novel Convolutional Neural Network, called Asymmetric Filtering-based Dense Convolutional Neural Network (AF D-CNN) to learn powerful features, which can extract different levels’ features and take advantage of identity information. Moreover, instead of using typical metric learning methods, we obtain the ranking lists by merging Joint Bayesian and re-ranking techniques which do not need dimensionality reduction. Finally, extensive experiments show that our proposed architecture performs well on four popular benchmark datasets (CUHK01, CUHK03, Market-1501, DukeMTMC-reID).

Integration Convolutional Neural Network for Person Re-Identification in Camera Networks

In this paper, we propose a novel deep model named integration convolutional neural network (ICNN) for person re-identification in camera networks, which jointly learns global and local features in a unified framework. To this end, the proposed ICNN simultaneously applies two kinds of loss functions. Specifically, we propose the soft triplet loss to learn global features which automatically adjusts the margin threshold within one batch. The soft triplet loss could alleviate the difficult in tuning parameters and therefore learns discriminative global features. In order to avoid the part misalignment problem, we learn latent local features by conducting local horizontal average pooling on the convolutional maps. Afterward, we implement the identification task on each local feature. We concatenate global and local features using a weighted strategy to present the pedestrian images. We evaluate the proposed ICNN on three large-scale databases. Our method achieves rank-1 accuracy of 92.13% on Market 1501, 61.4% on CUHK03 and 85.3% on DukeMTMC-reID, and the results outperform the state-of-the-art methods.

Deep feature learning with relative distance comparison for person re-identification

Identifying the same individual across different scenes is an important yet difficult task in intelligent video surveillance. Its main difficulty lies in how to preserve similarity of the same person against large appearance and structure variation while discriminating different individuals. In this paper, we present a scalable distance driven feature learning framework based on the deep neural network for person re-identification, and demonstrate its effectiveness to handle the existing challenges. Specifically, given the training images with the class labels (person IDs), we first produce a large number of triplet units, each of which contains three images, i.e. one person with a matched reference and a mismatched reference. Treating the units as the input, we build the convolutional neural network to generate the layered representations, and follow with the L2 distance metric. By means of parameter optimization, our framework tends to maximize the relative distance between the matched pair and the mismatched pair for each triplet unit. Moreover, a nontrivial issue arising with the framework is that the triplet organization cubically enlarges the number of training triplets, as one image can be involved into several triplet units. To overcome this problem, we develop an effective triplet generation scheme and an optimized gradient descent algorithm, making the computational load mainly depend on the number of original images instead of the number of triplets. On several challenging databases, our approach achieves very promising results and outperforms other state-of-the-art approaches.