search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.74k stars 7.96k forks source link

detect fire and smog select which one better between yolo_v3_tiny_pan_lstm.cfg.txt and yolo_v3_tiny_lstm? #4551

Open lyxbyr opened 4 years ago

lyxbyr commented 4 years ago

hello! i want to detect fire and smog based on video ,can distinguish them , now model can classify them use yolov3.cfg on my little dataset ,then i add Time and Space Information to train second model use 3DCNN in order to classify them better ,but I don not know how to take detect results into to the seconde model 's input(only save much frame result into the second model input?) ,because the seconde model 's input is 16frame or 32 frame or more. if only use darknet could you sugesstion to use which one .cfg .thanks @AlexeyAB @cenit i am sorry to @ you ,thanks you very much

AlexeyAB commented 4 years ago

If you have big objects and long time sequence (at least 8-16 frames), then use yolo_v3_tiny_lstm.cfg otherwise use yolo_v3_tiny_pan_lstm.cfg

blackwool commented 4 years ago

Hi @AlexeyAB thank you for great work ! I do not find yolo_v3_tiny_pan_lstm.cfg or yolo_v3_tiny_lstm.cfg file in current repository.Wwhere can i find it? also ,how to config conv-lstm layer in configure files, can you give an example ? thank you very much!