Table 3.

WCE or colonoscopy test detection results, or both (IOU > 0.5, batch-size 1).

Training Dataset	Methods	Testing Dataset	Backbone Network	Pre-Train	Input Size	Prec	Recall	F1 Score
WCE images	Hyb-SSDNet (ours)	WCE images	Inception v4	✓	$299\times 299$	93.29%(mAP)	89.4%(mAR)	91.5%(mAF)
ETIS-Larib+CVC-ClinicDB	Hyb-SSDNet (ours)	CVC-ClinicDB	Inception v4	✓	$299\times 299$	91.93%(mAP)	89.5%(mAR)	90.8%(mAF)
ETIS-Larib+CVC-ClinicDB	Hyb-SSDNet (ours)	ETIS-Larib	Inception v4	✓	$299\times 299$	91.10%(mAP)	87%(mAR)	89%(mAF)
WCE +CVC-ClinicDB	Souaidi et al., 2022 [18]	ETIS-Larib	VGG16	✓	$300\times 300$	90.02%(mAP)	×	×
CVC-ClinicDB + ETIS-Larib	Shin et al., 2018 [59]	ETIS-Larib	Inception ResNet	✓	$768\times 576$	92.2%	69.7%	79.4%
SUN+ PICCOLO+ CVC-ClinicDB	Ishak et al., 2021 [32]	ETIS-Larib	YOLOv3	✓	$448\times 448$	90.61%	91.04%	90.82%
CVC-ClinicDB	Liu et al., 2021 [60]	ETIS-Larib	ResNet-101	✓	$384\times 288$	77.80%	87.50%	82.40%
GIANA 2017	Wang et al., 2019 [61]	ETIS-Larib	AFP-Net(VGG16)	✓	$1225\times 996$	88.89%	80.7%	84.63%
CVC-ClinicDB	Qadir et al., 2021 [62]	ETIS-Larib	ResNet34	✓	$512\times 512$	86.54%	86.12%	86.33%
CVC-ClinicDB	Pacal and Karaboga, 2021 [63]	ETIS-Larib	CSPDarkNet53	✓	$384\times 288$	91.62%	82.55%	86.85%
CVC-ClinicDB	Wang et al., 2019 [61]	ETIS-Larib	Faster R-CNN (VGG16)	×	$224\times 224$	88.89%	80.77%	84.63%
CVC-VideoClinicDB	Krenzer et al., 2019 [64]	CVC-VideoClinicDB	YOLOv5	×	$574\times 500$	73.21%(mAP)	×	79.55%