Yolov2 is a good object detect framework. It has been introduced in research YOLO, YOLOv2, GitHub
There is a image detect using yolov2 framework
Using yolov2 to detect object, just modify main.py
args = dict({
# network toplogy and pre-training model
'cfg_file': 'files/yolo_yann.cfg',
'model_file': 'files/yolo_yann_89000.weights',
# test images directory
'test': 'test',
# output images directory
'output': 'output',
})
And then run main.py with
python main.py
To accelated train state, we use darknet(c++) to train our model, please see here. You should design your own model and after training, you can perform our framework to detect image, because it is more flexible than c++ version.