Hii mate thank u for sharing this implementation it's really nice as a beginner.
during my cracking to your code, I didn't get it, normally u output the model as one-hot encoded but in y_batch of the generator is label encoded
this is get_generator from BatchGenerator class
y_batch = np.zeros((self.config['BATCH_SIZE'], self.config['GRID_H'], self.config['GRID_W'], self.config['BOX'], 4+1+1)) # desired network output
this is the output of the model
Hii mate thank u for sharing this implementation it's really nice as a beginner.
during my cracking to your code, I didn't get it, normally u output the model as one-hot encoded but in y_batch of the generator is label encoded
this is get_generator from BatchGenerator class
y_batch = np.zeros((self.config['BATCH_SIZE'], self.config['GRID_H'], self.config['GRID_W'], self.config['BOX'], 4+1+1)) # desired network output
this is the output of the modeloutput = Reshape((GRID_H, GRID_W, BOX, 4 + 1 + CLASS))(x)
what do you think?