使用 google 物体识别的 api 来识别电影里面的物体。目前可以做到每秒 30 帧。
https://github.com/scotthuang1989/object_detection_with_tensorflow
分享自 TensorFlow QQ 群的彼岸：314148034
Target
detect people in a video or from camera. can choose what detection method we are using. Project Structure
From google
object label and detecion api data object_detection My module
myutil: help function object_detection_tf.py: main module for detecting object Usage
run following command python object_detection_tf.py -v /home/scott/Videos/S11E03.mp4
you need replace the video file with your choice Current Status
on my nvidia 1060 (6GB RAM ), the speed is ~25 frame per second, but gpu utilization is below 30%, I will use multiprocessing to speed up.
更多机器学习资源： http://www.tensorflownews.com/
Select Language