search

obendidi / Tracking-with-darkflow

Real-time people Multitracker using YOLO v2 and deep_sort with tensorflow
GNU General Public License v3.0
524 stars 174 forks source link

new trained tiny-yolo-voc model, track problem #28

Open zhangyujia0223 opened 6 years ago

zhangyujia0223 commented 6 years ago

I trained the new Yolo model. I just modified the following 2 aspects(filters and classes) in tiny-yolo-voc.cfg file as picture show. I want to track one of the classes we have labeled. image But there are some problems image Could you help me to solve this problem? Thank you very much.

zhangyujia0223 commented 6 years ago

When I test the video( just add the video file path), there are problems. I did not find your solution about it in issues. image Thanks again

obendidi commented 6 years ago

the last error is because your camera or video file is not working , or there is a typo the video path you provided , for the first you need to give more information , how did you generate the tiny-yolo-voc-p.weights ? after training it usually generates a .pb and .meta file ...

zhangyujia0223 commented 6 years ago

@bendidi

  1. For video, I have checked the video path, there is not a typo. I remember there are issues about solving this problem( you offered that: in run.py add print("Running for the first time") tfnet.camera("/home/ai2/yujia/data/videos/KT01.mp4"); in help.py uncomment the "file=self.FLAGS.demo" ), but now I can not find it, maybe it has been closed.
  2. we get the tiny-yolo-voc-p.weights from darknet, it does not generate .pb and .meta file.
zhangyujia0223 commented 6 years ago

@bendidi now I have solved the first one problem. But for the video, it still cannot be used. Could you give me guidance?

obendidi commented 6 years ago

as I've said, it's probably that your camera is not detected , try it alone to see if it is working or not , here is a little script to do that : http://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_gui/py_video_display/py_video_display.html

bit-scientist commented 5 years ago

@zhangyujia0223 Hi. It looks I need your help with my newly trained model. Let me give some info about it. I have trained a model for only one class using darknet yolo3 given here. Now I have a cfg file and .weights file. There are helmet_15000.cfg and helmet_15000.weights files. When I run I have similar error like your first image: `'Over-read {}'.format(self.path)

AssertionError: Over-read darkflow\bin\helmet_15000.weights` I know it's been long time since this post, but if you can help please do. Thanks.