I am using Caltech101 datasets from torchvision Pytorch library.
I am trying to build a subset of this large dataset with just five classes (Caltech5). As part of this, I am selecting few classes and segregating them based on the class index.
However, I am unable to map class index given by Caltech101 dataloader to the list of class label correctly.
Class labels: caltech101_class_labels.txt
The very first image given by the dataloader itself is not matching.
dataloader = datasets.Caltech101(root="./", download=True) idx = 0 img, _class = dataloader[idx] print("Class: ", _class) plt.imshow(img) plt.show()
Files already downloaded and verified
Class index 0 should belong to accordion but it is a face.
Am i missing anything here ?