llSourcell / YOLO_Object_Detection

This is the code for "YOLO Object Detection" by Siraj Raval on Youtube
GNU General Public License v3.0
1.73k stars 795 forks source link

Training on your own dataset #16

Open tahanakabi opened 6 years ago

tahanakabi commented 6 years ago

I have red the part where it's explained how to train the model using your own dataset. However, it doesn't tell how and where to put the training images. There is a mention of the dataset here: flow --model cfg/tiny-yolo-voc-3c.cfg --load bin/tiny-yolo-voc.weights --train --annotation train/Annotations --dataset train/Images but I don't know how the images should be organized with labels. My training dataset is organized in folders each folder is named after the class of the images that it contains. Is there anyway I can use this code on such training dataset?

Thank you!

saratrajput commented 5 years ago

You are referring to weakly supervised object detection, where you localize objects in your image based on only image level labels. I suggest you look into that. This repo is useful if you are willing to hand-annotate your data with bounding boxes.