Table 4 Comparisons of detection performance and inference speed for YOLOV4-tiny-3Â l by using the proposed and baseline approach on model development test set data
From: Tile-based microscopic image processing for malaria screening using a deep learning approach
Model | Tile size | AP (%) | R (%) | Time (sec/img) | |
|---|---|---|---|---|---|
Train | Inference | ||||
YOLOV4-tiny-3l@416 | 1088 \(\times\) 1088 | 1088 \(\times\) 1088 | 55.4 | 59.8 | 1.0 |
832 \(\times\) 832 | 65.9 | 71.2 | 1.0 | ||
608 \(\times\) 608 | 76.9 | 84.6 | 1.4 | ||
YOLOV4-tiny-3l@512 | 1088 \(\times\) 1088 | 1088 \(\times\) 1088 | 67.2 | 72.8 | 1.3 |
832 \(\times\) 832 | 78.0 | 84.3 | 1.4 | ||
608 \(\times\) 608 | 83.4 | 92.3 | 2.0 | ||
YOLOV4-tiny-3l@608 | 1088 \(\times\) 1088 | 1088 \(\times\) 1088 | 78.4 | 84.3 | 1.2 |
832 \(\times\) 832 | 86.1 | 93.0 | 1.6 | ||
608 \(\times\) 608 | 81.6 | 89.3 | 2.0 | ||
YOLOV4-tiny-3l@416 | 832 \(\times\) 832 | 1088 \(\times\) 1088 | 50.0 | 53.7 | 1.0 |
832 \(\times\) 832 | 64.3 | 70.0 | 1.0 | ||
608 \(\times\) 608 | 80.5 | 87.7 | 1.5 | ||
YOLOV4-tiny-3l@512 | 832 \(\times\) 832 | 1088 \(\times\) 1088 | 64.9 | 71.4 | 1.1 |
832 \(\times\) 832 | 78.1 | 84.5 | 1.4 | ||
608 \(\times\) 608 | 85.7 | 94.2 | 1.9 | ||
YOLOV4-tiny-3l@608 | 832 \(\times\) 832 | 1088 \(\times\) 1088 | 74.5 | 81.0 | 1.3 |
832 \(\times\) 832 | 85.4 | 92.6 | 1.6 | ||
608 \(\times\) 608 | 84.0 | 91.0 | 2.0 | ||
YOLOV4-tiny-3l@416 | 608 \(\times\) 608 | 1088 \(\times\) 1088 | 41.8 | 57.1 | 0.8 |
832 \(\times\) 832 | 64.4 | 71.4 | 1.0 | ||
608 \(\times\) 608 | 81.7 | 89.7 | 1.4 | ||
YOLOV4-tiny-3l@512 | 608 \(\times\) 608 | 1088 \(\times\) 1088 | 51.8 | 57.2 | 1.0 |
832 \(\times\) 832 | 74.6 | 81.1 | 1.3 | ||
608 \(\times\) 608 | 87.1 | 95.0 | 1.9 | ||
YOLOV4-tiny-3l@608 | 608 \(\times\) 608 | 1088 \(\times\) 1088 | 69.2 | 85.7 | 1.0 |
832 \(\times\) 832 | 83.8 | 91.9 | 1.4 | ||
608 \(\times\) 608 | 87.4 | 95.1 | 1.9 | ||
Without tiling | |||||
YoloV4-tiny-3l@416 | – | – | 71.46 | 71.0 | 0.2 |
YoloV4-tiny-3l@512 | – | – | 79.4 | 78.0 | 0.2 |
YoloV4-tiny-3l@608 | – | – | 78.73 | 76.0 | 0.2 |