Mobile-YOLO

A mobile helmet detection method. The method is mainly based on the MobileNetv2 and YOLOv4

The project utilizes the Darknet from https://github.com/AlexeyAB/darknet

Dir cfg contains the configuration files

Dir weights contains the weight files

Dir mobile contains configuration and weight file after weight quantization using NCNN https://github.com/Tencent/ncnn

Results

The model is trained on VOC2007+2012 train&val and tested on VOC2007 test dataset with an mAP of 73.1. Results on private helmet dataset is 89.3%.

To begin

PCs

Please follow https://github.com/AlexeyAB/darknet to setup. Replace the .cfg and .weights files with our files.

Mobile Devices

Please follow https://github.com/Tencent/ncnn to setup and use our .param and .bin files.