Object Detection on HiXray (test)
91.2PrecisionYOLOv8m
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| YOLOv8mModel Size=50.3MB, GFLOPs=79.1G, FPS=96.23, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 91.2 | 80.6 | 83.9 | 55 | |
| GSA-YOLOModel Size=5.7MB, GFLOPs=8.0G, FPS=189.62, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 90.2 | 79.5 | 82.7 | 53.1 | |
| GEMA-YOLOModel Size=11.8MB, GFLOPs=16.1G, FPS=82.57, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 89.5 | 78.8 | 82.2 | 52.3 | |
| DSF-YOLOModel Size=35.4MB, GFLOPs=43.4G, FPS=54.60, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 88.9 | 78.6 | 81.7 | 51.5 | |
| ESI-YOLOModel Size=6.7MB, GFLOPs=8.9G, FPS=164.48, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 88.6 | 77.9 | 81.5 | 51.7 | |
| SRW-YOLOModel Size=9.4MB, GFLOPs=12.0G, FPS=158.97, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 88.3 | 78.2 | 81.4 | 51.6 | |
| YOLOv8nModel Size=6.2MB, GFLOPs=8.7G, FPS=170.58, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 88.2 | 76.9 | 80.6 | 50.7 | |
| Fine-YOLOModel Size=46.2MB, GFLOPs=56.9G, FPS=152.42, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 87.7 | 77.1 | 81.1 | 51.4 | |
| YOLOv8n-Pruned (L1)Model Size=4.4MB, GFLOPs=6.1G, FPS=231.92, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point), Pruning=L12026.05 | 86.9 | 76 | 79.5 | 49.6 | |
| YOLOv5nModel Size=3.9MB, GFLOPs=4.5G, FPS=161.31, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 85.9 | 77.3 | 79.9 | 48.6 | |
| YOLOv8n-Pruned+QAT (INT8)Model Size=2.3MB, GFLOPs=6.1G, FPS=312.50, Input Resolution=640 × 640, Measurement Precision=INT8, Quantization=QAT2026.05 | 85.6 | 74.8 | 78.2 | 48.7 | |
| Faster R-CNNModel Size=319.2MB, GFLOPs=168.7G, FPS=15.02, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 84.1 | 76.3 | 78.8 | 46.7 | |
| RetinaNetModel Size=128.7MB, GFLOPs=151.5G, FPS=34.05, Input Resolution=640 × 640, Measurement Precision=FP32 (32-bit floating-point)2026.05 | 82.4 | 72.3 | 75.6 | 45.1 |