(14/11/2022 - 20/11/2022)
Used the CNN model testing for the cropped drone images.
Best pre trained architecture - VGG16
Train the model using different amount of data set using VGG16 and observed the training and testing accuracy.
VGG 16Â codes
Searched other codes than previous one for shadow detection.
Because, the previous code didn't recognize the shadow in all drone images. It recognized incorrect shadow edge.