A Vehicle Detection Model from Still Images using Modified RT-DETR with Channel and Spatial Attention Modules

Abstract

For developing an automated traffic management system, it would be highly advantageous to design automatic vehicle detection (AVD) systems that use either images or videos from traffic data. The goal of this paper is to develop a reliable vehicle detection model that can detect different kinds of vehicle in a variety of lighting and weather conditions, such as sunny, rainy, daytime and nighttime. We have designed our model using a modified RT-DETR (Real-Time Detection Transformer), which is a basically Transformer-based object detection model pre-trained on the COCO dataset. We have modified this by adding the Channel Attention and Spatial Attention modules sequentially. For the experimental evaluation, we have tested our model on two recently introduced vehicle detection datasets, namely JUVDsiv1 and IRUVD. We have achieved a satisfactory result on these two datasets. We have obtained a 77% mAP score on the JUVDsiv1 dataset, and 96.3% mAP score on the IRUVD dataset.

More by Dmitry Kaplun

• Codewithsayanjib/slm-systematic-review: v1.1Zenodo · 2026
• Microscopy Image Dataset of pulmonary vessels for Quantitative assessment of fibrosisScientific Data · 2026
• Microscopy Image Dataset of pulmonary vessels for Deep Learning-Based Quantitative assessment of fibrosisFigshare · 2026
• MA2MU-Net++: A Mahalanobis distance and Ackley function based multi-modal U-Net++ model for breast nuclei segmentation in histopathology imagesBiomedical Signal Processing and Control · 2026