YOLO v2
YOLO9000 is a high speed, real time detection algorithm that can detect on OVER 9000! (object categories). Read more
Requirements
- Python 3.5 or 3.6. Download here
- Tensorflow. Download here
- OpenCV Download here
pip install opencv-python
Download Darkflow repo
- Download ZIP
- Extract.
Build the library
- Cd to
darkflow-master
folder that just extracted. - Open terminal & Run:
python setup.py build_ext --inplace
or
pip install -e .
Download the 'weights' file
- Create
bin
folder insidedarkflow-master
folder. - Download the YOLOv2 608x608 weights file here.
- Put the
weights
file insidebin
folder.
Processing video file
- Put the video into
darkflow-master
folder. - Open terminal & Run:
- Without GPU version of tensorflow:
python flow --model cfg/yolo.cfg --load bin/yolov2.weights --demo video_file_name.mp4 --saveVideo
- GPU version of tensorflow:
python flow --model cfg/yolo.cfg --load bin/yolov2.weights --demo video_file_name.mp4 --gpu 1.0 --saveVideo
*NOTE:
-
video_file_name.mp4 changes to your video name.
-
--saveVideo indicates to save a name video file, which has the boxes around objects