Object Detection on PCB surface-defect dataset benchmark leaderboard · SOTA2 Research

Object Detection on PCB surface-defect dataset

95.84mAP@0.5YOLOX-S

Evaluation Results

Method	Links
YOLOX-S2026.03		95.84	62.03	87.16	93.73	91
YOLOR-P62026.03		94.7	54.5	93.1	93.1	94
DS-MoE2026.03		94.24	53.36	83.38	96.8	89
AirDet-S2026.03		94.1	51.81	82.91	95.45	89
PP-YOLOE-S2026.03		93.44	53.11	81.4	93.31	87
YOLOv82026.03		93.25	51.35	81.93	95.21	86
MLLA2026.03		93.19	51.71	82.74	95.18	86
YOLOv72026.03		93.11	52.39	85.12	94.68	89
PromptDet2026.03		93.1	51.8	82.9	94.5	89
SegMAN2026.03		92.21	50.46	80.32	95.52	88
MViTv2-B2026.03		92.13	51.56	81.21	95.44	89
YOLOv5-X2026.03		91.8	50.7	95.2	91.3	93
MiniGPT-V2026.03		91.5	50.1	82.3	93.1	89
YOLOv6-S2026.03		91.42	49.44	84.12	95.31	89
YOLOv5-S2026.03		91.2	51.8	81.82	92.12	86
YOLOv42026.03		81.2	50.06	89.37	94.78	92
Faster-R-CNN2026.03		78.33	50.24	67.74	74.51	70
YOLOv32026.03		73.32	47.5	90.85	52.66	67
EfficientDet-D32026.03		71.8	35	99.44	49.51	66
CenterNet2026.03		43.4	14.53	43.79	52.2	48
SSD2026.03		17.1	10	19.93	54.64	29
RetinaNet2026.03		13.18	5	66.67	4.31	8