Transfer Learning for Custom Object Detection

WongKinYiu / yolov7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors

GNU General Public License v3.0

13.26k stars 4.18k forks source link

Transfer Learning for Custom Object Detection #2013

Open rut00 opened 5 months ago

rut00 commented 5 months ago

Greetings,

I'm attempting transfer learning on the YOLOv7 model to include custom objects alongside the existing 80 classes. Despite retraining the pre-trained model, the original classes reset post-training. I have also tried freezing the layers, and modifying config files, and hyperparameter files but I am not getting hold of pre-trained classes.

Please provide insight for the same.

dsbyprateekg commented 5 months ago

The way you are following does not work with YOLO. You must add your new class data to the existing 80 classes dataset and train the model if you want to detect the newly added custom class with the existing 80 classes.