In some cases, I'd like to load the dataset after some transformations and do some analysis on the datapoint before feeding them to the (neural net) training, and I'd like to do this analysis on the raw image, rather than let's say a randomly cropped version. But after doing the analysis, I'd like to select a subsample of the dataset and train the dataset using them. This would be the case in many applications such as Active Learning. Is there any easy way to have this kind of functionality when using FCCV Loaders?
Hello, thanks for the great work.
In some cases, I'd like to load the dataset after some transformations and do some analysis on the datapoint before feeding them to the (neural net) training, and I'd like to do this analysis on the raw image, rather than let's say a randomly cropped version. But after doing the analysis, I'd like to select a subsample of the dataset and train the dataset using them. This would be the case in many applications such as Active Learning. Is there any easy way to have this kind of functionality when using FCCV Loaders?