Multi class object detection loading label csv and data


I have a training set of 320 pictures of cubes that are red, green and blue. Some pictures have one cube in it, some have multiple cubes in it. I compiled a training labels csv file that has a line for each object that was labeled in a picture:

How can I properly load the data in a databunch to start training a model to recognize and localize the cubes?

Thanks in advance for the help!

PS: I know the image size is ridiculous, but this was my first attempt ever :stuck_out_tongue: