Image-Based Person Re-identification Using Omni-Scale Feature Learning

Nowadays, person re-identification (re-id) has received a large interest from the research community. It has been considered the heart of the studies of computer vision that plays a vital role in security applications to prevent terrorist attacks and track robberies. This paper highlights the design of the recently presented network Omni-Scale Network in image-based person re-id which has fewer parameters than the best-performing Residual-based Network methods. Then, views some datasets that have been used with such a network as well as our custom-created dataset which is better in performance than some well-known small datasets. The custom-created dataset contains 256 images collected using mobile phone cameras. The results of the different datasets applied to the proposed network are compared and evaluated.