rafaelpadilla / darknet

Useful functionalities added on the original darknet public repository.
Other
36 stars 17 forks source link

can i use yolo v3 with your modification? #1

Closed yojun0611 closed 6 years ago

yojun0611 commented 6 years ago

hello, rafaelpadilla first of all, thank you for your work because it's exactly what I want. I am very new in this field. so I have little knowledge about computer science.

I want to get objected bounding box and re-label to re-train model. As my first step, I need objected bounding box for re-labeling to re-train model. (which exactly your work provides. thx again)

so I did 'git clone' your repo. and I modified Makefile from GPU=0 to 1 for using GPU. and I did 'make'. then I did do your example.

It did look like working properly. However, it couldn't detect any objects like below. default can u tell me what did I wrong?

and next question is the same as the title. I tried to do your example with yolov3.cfg & weight. then I got an error like below. image is there any way to use yolo v3 with your modification?

I hope that you could help me. Thank you a lot again anyway.

rafaelpadilla commented 6 years ago

Dear @yojun0611,

First of all, sorry for the delay on answering your message. Somehow I am not being alerted when an issues is opened.

Answering your questions:

1) As the weight file distributed in this repository was about 268MB, I had to commit this file with git-lfs (Large File Storage). It will not work if you just download the project. Please, try cloning the project instead of downloading it. Make sure the yolo-voc_final.weights is about 268MB. If not, download it manually from the repo. Please, let me know if you got it working.

2) No, this repository is for YOLO V2. Soon I will apply these same features (testing multiple images, testing with different threshold values, etc) with YOLO V3. I believe that if you click on the "start" button, you will be notified with the new changes.

Thanks a lot for your feedback!