YOLOv7-Tiny and YOLOv8n Evaluation for Face Detection

The lightweight face detection models were developed to match the specifications of edge devices. This makes them suitable for use in real-life applications where the high GPU might not be available. This study compares the performance of two lightweight object detection models, YOLOv7-tiny and YOLOv8n, for face detection applications. Both models were trained on WIDERFACE dataset, and their performance was evaluated using mean average precision (mAP). Regarding model size, YOLOv8n is lighter than YOLOv7-tiny with only 3.01 million parameters, making it more efficient for deployment on resource-constrained devices. In terms of accuracy, the results showed that YOLOv7-tiny achieved a mAP50 of 69.8% and a mAP50-95 of 36.2%, while YOLOv8n achieved a mAP50 of 37% and an mAP50-95 of 67.6%