Real time detection of Koala bear with finetuned YOLOv3 network initialized with pretrained weights on COCO dataset
-
Create Dataset folder and include following items in the folder a) test.txt - Contains list of all test images b) train.txt - contains all training images c) All training and testing images in .jpg format with labels from labellmg tool d) classes.names - contains all class names e) data_locations.data - contains all location information regarding data and backup
-
Download Pretrained weights on COCO dataset from https://pjreddie.com/media/files/darknet53.conv.74
-
Modify yolov3.cfg file according to the number of classes
-
Training the Model: train.ipynb
-
Testing the Model: ''' python3 koala_Real_time_detection.py '''