##yolov3
Step1：convert official weights to keras compatible type(.h5)
download yolov3.weights from https://pjreddie.com/media/files/yolov3.weights
download yolov3 config file from https://github.com/pjreddie/darknet/blob/master/cfg/yolov3.cfg
use convert.py to convert weights format
tools/convert.py path/to/yolov3.cfg path/to/yolov3.weights outpath/to/yolov3.h5
Step2:
Try some classifiction task.
run trains/train_classification.py with classification config.
In this experiment, Oxford flowers dataset is used. data can be downloaded here https://download.csdn.net/download/tsyccnh/10641502
run predicts/flowers_predict.py to evluate on test set.
The test accuracy should be around 97%
Step3:
...to be added