search

pjreddie / darknet

Convolutional Neural Networks
http://pjreddie.com/darknet/
Other
25.88k stars 21.33k forks source link

Different results for video and image analysis #2012

Open nsivaramakrishnan opened 4 years ago

nsivaramakrishnan commented 4 years ago

Hi, We are trying to YoloV3 tiny yolo (V2) model for object detection. We have our custom model trained on the objects. When we run the model on video file, the detections are lesser than the detections that are observed when run on the image file (image file is dumped using save_image API on the same video). Is there anything that we are missing for video? Thanks -Siva

GursimranSe commented 4 years ago

hi i run it on colab when i use video for detection !./darknet detector demo cfg/coco.data cfg/yolov3.cfg yolov3.weights data/YouTube.mp4 after run it video not showing or playing any idea how to fix it

nsivaramakrishnan commented 4 years ago

Hi, Is detection working in case of image? If yes, can you reduce the threshold and check if anything is getting detected?

hi i run it on colab when i use video for detection !./darknet detector demo cfg/coco.data cfg/yolov3.cfg yolov3.weights data/YouTube.mp4 after run it video not showing or playing any idea how to fix it

Ella2le commented 4 years ago

I have the same problem, the model can detect all objects on the same frame, but when detecting video the [demo] can only detect single or noting. have you solved it?