Home

MobileNet-YOLO

The purpose of this project was provide a platform which can change backbone easily. So I select the caffe framework and write down my training history here.

Enjoy it

Build

1. Install dependency
2. Following below command

> git clone https://github.com/eric612/MobileNet-YOLO.git 
> cd $MobileNet-YOLO_root/
> mkdir build`
> cd build
> cmake ..
> make -j4
> make pycaffe

Run demo

> cd $caffe_root/
> sh demo_yolo_lite.sh

Training

Download lmdb

Unzip into $caffe_root/

Please check the path exist

"$caffe_root\examples\VOC0712\VOC0712_trainval_lmdb"

"$caffe_root\examples\VOC0712\VOC0712_test_lmdb"

Download pre-trained weights, and save at $caffe_root\models\MobileNet

> cd $caffe_root/
> sh train_yolov3_lite.sh

Train your own dataset

How to create lmdb

Parameter-tuning

Generate anchors to increase performance

An example python code to generate yolo prototxt

YOLOv3-Tiny

See issue #16

Edge computing

ncnn,tensortRT,jetson tx 2 , ... , my collection link

Issue

Here has some issue which I'm not sure my implement was correct , if you have answers , please let me know

1. Multiple resolution training : inference at resolution 608 do not get good accuracy
2. Prefetch problem : In order to implement multiple resolution training , I fixed the prefetch count = 1, this may decrease training efficiency and make some bugs when use multiple gpu. (If I have time , I will fix this bugs)