An optimised YOLOv4 deep learning model for efficient malarial cell detection in thin blood smear images

Parasites & Vectors

Table 4 Network structure of YOLOv4. Network structure of the YOLOv4 model

YOLOv4 mels	Number of layers	Type
CSPDarkNet53 without the fully connected layer	0	CBL
	1–7	Res
	8	Conv
	9	Route
	10–20	2Res
	21	Conv
	22	Route
	23–51	8Res
	52	Conv
	53	Route
	54–82	8Res
	83	Conv
	84	Route
	85–101	4Res
	102	Conv
	103	Route
Feature fusion layer and output layer	104–107	4CBL
	108–113	SPP
	114–117	4CBL
	118	Up-sample
	119	Route
	120	Conv
	121	Route
	122–127	6CBL
	128	Up-sample
	129	Route
	130	CBL
	131	Route
	132–137	6CBL
	138–139	CBL + YOLO
	140	Route
	141	CBL
	142	Route
	143–148	6CBL
	149–150	Conv + YOLO
	151	Route
	152	CBL
	153	Route
	154–159	6CBL
	160–161	Conv + YOLO

Head: The main function is to locate the bounding boxes and classify the objects of interest. The coordinates and the scores of every bounding box are generated
YOLO You Only Look Once (model), CBL Convolutional, Batch normalisation, and Leaky-ReLU (Feature extractor), Res residual block

ISSN: 1756-3305