We are going to train custom image classification models using transfer learning technique and ConvNets (Convolutional Neural Nets). Note that there is a parameter for undersampling. So you don't need to under sample manually before running the script. It is passed to the RandomUnderSampler class within imblearn library. You'll probably need to change some parameters here, depending on the complexity of your dataset and its labels.
Dec-26-2021, 00:30:44 GMT