Part-based Signal Modulation and Adaptive Skip Merger for Person Search

Document Type



Person search is a computer vision problem that joins the two tasks of person detection and person re-identification (ReID). Previous works handle this problem with either a two-step or end-to-end approach, and has attained much attention due to complex challenges in the scene such as appearance variations, background clutter, and deformation. These approaches achieve significant performance but are still prone to performance degradation under complex scenes which may jeopardize the accuracy of person search methods. In this thesis, we propose a novel part-based signal modulation module for person search (PSM-PS) within a Faster R-CNN based person search framework. The proposed PSM module transforms the person parts, represented as part tokens, in a wave-like manner, where amplitude indicates the real part and phase shows the imaginary part in a complex domain. PSM modulates the pedestrian part tokens such that it enhances the feature representation where the close parts of the person have a close phase compared to others. Furthermore, we propose a second novel adaptive skip merger module for person search (ASM-PS), also within a Faster R-CNN based person search framework. The ASM module achieves enriched aggregated features through element-wise merged weight assignment to allow emphasis on the most significant features. ASM highlights the most informative regions through merging high and low-level features and determining the relevance of the input features. Moreover, The experiments are performed over the two prominent person search datasets: CUHK-SYSU and PRW. The extensive experimental study demonstrates the effectiveness of our methods and shows the state-of-the-art performance compared to other methods.

First Page


Last Page


Publication Date



Thesis submitted to the Deanship of Graduate and Postdoctoral Studies

In partial fulfillment of the requirements for the M.Sc degree in Computer Vision

Advisors: Dr. Rao Anwer

Online access for MBZUAI patrons