{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/vehicle-re-identification-using-quadruple","title":"Vehicle Re-identification Using Quadruple Directional Deep Learning Features","arxiv_id":"1811.05163","date":"2018-11-13","proceeding":null,"authors":["Jianqing Zhu","Huanqiang Zeng","Jingchang Huang","Shengcai Liao","Zhen Lei","Canhui Cai","Lixin Zheng"],"abstract":"In order to resist the adverse effect of viewpoint variations for improving\nvehicle re-identification performance, we design quadruple directional deep\nlearning networks to extract quadruple directional deep learning features\n(QD-DLF) of vehicle images. The quadruple directional deep learning networks\nare with similar overall architecture, including the same basic deep learning\narchitecture but different directional feature pooling layers. Specifically,\nthe same basic deep learning architecture is a shortly and densely connected\nconvolutional neural network to extract basic feature maps of an input square\nvehicle image in the first stage. Then, the quadruple directional deep learning\nnetworks utilize different directional pooling layers, i.e., horizontal average\npooling (HAP) layer, vertical average pooling (VAP) layer, diagonal average\npooling (DAP) layer and anti-diagonal average pooling (AAP) layer, to compress\nthe basic feature maps into horizontal, vertical, diagonal and anti-diagonal\ndirectional feature maps, respectively.\n  Finally, these directional feature maps are spatially normalized and\nconcatenated together as a quadruple directional deep learning feature for\nvehicle re-identification. Extensive experiments on both VeRi and VehicleID\ndatabases show that the proposed QD-DLF approach outperforms multiple\nstate-of-the-art vehicle re-identification methods.","url_abs":"http://arxiv.org/abs/1811.05163v1","url_pdf":"http://arxiv.org/pdf/1811.05163v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"vehicle-re-identification","task_name":"Vehicle Re-Identification"}],"methods":[{"method_slug":"average-pooling","method_name":"Average Pooling"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/vehicle-re-identification-on-veri-776","task":"Vehicle Re-Identification","dataset":"VeRi-776","model":"QD-DLF","rank_in_archive_order":16,"of":17,"metrics":{"mAP":"61.83"},"uses_additional_data":false},{"leaderboard":"/sota/vehicle-re-identification-on-vehicleid-large","task":"Vehicle Re-Identification","dataset":"VehicleID Large","model":"QD-DLF","rank_in_archive_order":10,"of":10,"metrics":{"mAP":"68.41"},"uses_additional_data":false},{"leaderboard":"/sota/vehicle-re-identification-on-vehicleid-medium","task":"Vehicle Re-Identification","dataset":"VehicleID Medium","model":"QD-DLF","rank_in_archive_order":9,"of":9,"metrics":{"mAP":"74.63"},"uses_additional_data":false},{"leaderboard":"/sota/vehicle-re-identification-on-vehicleid-small","task":"Vehicle Re-Identification","dataset":"VehicleID Small","model":"QD-DLF","rank_in_archive_order":13,"of":13,"metrics":{"mAP":"76.54"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.05163","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}