Hello good people.
So I would want to predict milk yield from image data. I have a sample of 1000 images. Would this suffice to train my model? Also, how much can I augment the data so as to create more samples.
If you know of published literature on these issues, please provide a link.
1,000 images is not a very large data set but you should try and see if the model will be able to show reasonable accuracy. The important thing is to set aside a small portion of this data set for validation and check the accuracy on the images that were not used for training.
You can use data augmentation methods as is demonstrated in this tutorial: Data augmentation | TensorFlow Core
Random horizontal flips, random changes in brightness and contrast and some other transformations could be safely used to make your data set more diverse.