yolov3-ios
Using yolo v3 object detection on ios platform.
Example applications:
QuickStart:
Run tiny_model.xcodeproj in ios.
Training
The training process mainly consults qqwweee/keras-yolo3. We add yolov3 with Densnet.
1.Requirement
python 3.6.4
keras 2.1.5
tensorflow 1.6.0
2.Generate datasets
Generate datasets with VOC format. And try python voc_annotations
.
3.Start training
For yolo model with darknet:
wget https://pjreddie.com/media/files/darknet53.conv.74
rename it as darknet53.weights
python convert.py -w darknet53.cfg darknet53.weights model_data/darknet53_weights.h5
python yolov3_train.py
, with model_data/darknet53_weights.h5 as pre-trained model
For yolo model with densenet:
Converting
1.Building environment
virtualenv -p /usr/bin/python2.7 keras_coreml_virt
source keras_coreml_virt/bin/activate
pip install protobuf
pip install tensorflow==1.6.0
pip install keras==2.1.5
pip install h5py
pip install coremltools==0.8.0
2.Convert .h5 model to .mlmodel
python coreml.py
Building project in Xcode
For yolo model with darknet or densenet
For tiny model