Table 1.
Fig 1.
Abstract-level overview of ECE-VDTDA system.
Fig 2.
Applications of vehicle detection.
Fig 3.
Research methodology of ECE-VDTDA system.
Table 2.
SimYOLO-V5s_WIOU anchor boxes grouped by detection layer, stride, and object size (input size 640×640).
Fig 4.
Architectural diagram of SimYOLO-V5s_WIOU vehicle detection algorithm.
Fig 5.
Bounding boxes and intersection over union with central points (Red) and smallest enclosing box (blue) [22].
Fig 6.
Vehicle detection and tracking in the ECE-VDTDA system.
Fig 7.
Flow chart of the ECE-VDTDA system.
Table 3.
Threshold levels and warnings/alerts for vehicle speed, distance, and TTC estimations in the ECE-VDTDA system.
Fig 8.
FD, DAWN, and FC datasets class-wise labels distribution.
Table 4.
Comparison of vehicle detection performance (accuracy) of baseline YOLO-V5s and SimYOLO-V5s_WIOU on the FD, DAWN, and FC datasets.
Table 5.
Comparison of vehicle detection performance (speed) of the baseline YOLO-V5s and SimYOLO-V5s_WIOU on the FD, DAWN, and FC datasets.
Table 6.
Comparison of vehicle detection performance of SOTA and SimYOLO-V5s_WIOU on FD dataset.
Table 7.
Comparison of vehicle detection performance (speed) of SimYOLO-V5s_WIOU with SOTA on the FD dataset.
Table 8.
Comparison of vehicle detection performance of state-of-the-art and SimYOLO-V5s_WIOU on DAWN dataset.
Table 9.
Comparison of vehicle detection performance (speed) of SimYOLO-V5s_WIOU with state-of-the-art on the DAWN dataset.
Fig 9.
Visual comparison of vehicle detection performance of baseline YOLO-V5s and optimized SimYOLO-V5s_WIOU on set-I.
Fig 10.
Visual comparison of vehicle detection performance of baseline YOLO-V5s and optimized SimYOLO-V5s_WIOU on set-II.
Fig 11.
Loss, P, R, and mAP graphs of SimYOLO-V5s_WIOU on FD, DAWN, FC datasets.
Table 10.
Comparison of vehicle detection performance (accuracy) of SimYOLO-V5s_WIOU with state-of-the-art SimYOLO-V5s variants on the FC dataset.
Table 11.
Comparison of vehicle detection performance (speed) of SimYOLO-V5s_WIOU with state-of-the-art SimYOLO-V5s variants on the FC dataset.
Table 12.
Comparison of vehicle detection and tracking performance of the ECE-VDT system with SOTA on the BDD100K video sequence of 1213 frames.
Table 13.
Comparison of vehicle detection and tracking performance of the ECE-VDT system with SOTA on the BDD100K video sequence of 1213 frames.
Table 14.
Comparison of vehicle detection and tracking performance of the ECE-VDT system with SOTA on the self video sequence of 10213 frames.
Fig 12.
Visualization of Vehicle detection and tracking performance of ECE-VDTDA system.
Fig 13.
Comparative analysis of processing time, FPS, distance(m)-TTC(s), and speed(km/h)-TTC(s) alerts on web, BDD100K, and self-collected diverse weather datasets for collision avoidance and driver assistance.
Fig 14.
Visual comparison of distance(m)-TTC(s) and speed(km/h)-TTC estimations and alerts on web, BDD100K, and self-collected diverse weather datasets for collision avoidance and driver assistance.