Multi-scale Deep Learning Architectures for Person Re-identification

Abstract

Person Re-identification (re-id) aims to match people across non-overlapping camera views in a public space. It is a challenging problem because many people captured in surveillance videos wear similar clothes. Consequently, the differences in their appearance are often subtle and only detectable at the right location and scales. Existing re-id models, particularly the recently proposed deep learning based ones match people at a single scale. In contrast, in this paper, a novel multi-scale deep learning model is proposed. Our model is able to learn deep discriminative feature representations at different scales and automatically determine the most suitable scales for matching. The importance of different spatial locations for extracting discriminative features is also learned explicitly. Experiments are carried out to demonstrate that the proposed model outperforms the state-of-the art on a number of benchmarks.

Omni-Scale Feature Learning for Person Re-Identification

“…Overall, the strong performance of OSNet on these two small datasets is indicative of its practical usefulness in real-world applications where collecting large-scale training data is unscalable. [34]. As a result, the R1/mAP drops off by 2.0%/3.5% compared with that of dynamic gates (primary model).…”

“…Nonetheless, the importance of multi-scale feature learning has been recognised recently and the multi-stream building block design has also been adopted. Compared to a number of ReID networks with multi-stream building blocks [2,34], OSNet is significantly different. Specifically the layer design in [2] is based on ResNeXt [62], where each stream learns features at the same scale, while our streams in each block have different scales.…”

“…In contrast, our building block uses a scale-controlling factor to diversify the spatial scales to be captured. Moreover, [34] fuses multi-stream features with learnable but fixed-oncelearned streamwise weights only at the final block. Whereas we fuse multi-scale features within each building block using dynamic (input-dependent) channel-wise weights to learn combinations of multi-scale patterns.…”

Spectral Feature Transformation for Person Re-Identification

“…One is to improve the network architecture for person ReID. Besides common techniques in CNN such as multi-scale feature aggregation [27] or attention modules [18,44], tailor-made architectures [41,39,30,49] for person ReID are also devised. Sun et al [41] splited the feature map into several horizontal parts and imposed supervision on them directly.…”

Video-Based Convolutional Attention for Person Re-Identification

“…In fact, CNNs are able to extract different features from a given image, representing them as a set of output maps avoiding manual effort in fea-ture engineering. Image-based Automatic Person Re-Identification is one of the fields in which CNNs achieved remarkable results [19,20,21,22,23,24].…”