YoloV3 Simplified for training on Colab with custom dataset for one class (GUN)
Class - gun
- We have added a 500 images of unique object (gun) in the folder customdata after annotating the images using Annotation Tool. The structure we followed to store them is
data
--customdata
--images/
--img001.jpg
--img002.jpg
--...
--labels/
--img001.txt
--img002.txt
--...
custom.data #data file
custom.names #class name
customtrain.txt #list of name of the images to train our network.
customtest.txt #list of names of the images for validation
- For one class example our custom.data is here. We used 500 images for training and 100 images for testng.
- downloaded the weights (yolov3-spp-ultralytics.pt) from the original and placed in Google Drive.
- Created a weights folder under YoloV3 to store weights
- Trained for 300 epochs after configuring. (log)[https://github.com/sridevibonthu/YoloV3/blob/master/results.txt)
Results
After training for 300 Epochs, results look awesome!
Performance