A lot of websites on CNN for large datasets of images talk about starting with the pretrained model for 1.2 million images in 1000 categories available via AlexNet / Imagenet. These sites seem to imply that this dataset is freely available, but I'm having trouble actually getting access to it. For example, I tried going to https://github.com/deep-diver/AlexNet but couldn't get the code in alexnet.ipynb to run. Consider the following code:

for f in data_file.iterdir():
    data = pickle.loads(f.read_bytes(), encoding='bytes')
    if 'meta' in str(f):
        meta = data
    elif 'test' in str(f):
        test = data
    elif 'batch' in str(f.name):
        train_dic = merge(train_dic, data)

This seems to imply that there should be some data files which include the text strings 'meta', 'test', and/or 'batch', but I couldn't seem to find those files.

Where is the best place to get access to this dataset--ideally somewhere where there is good documentation as to how to download/use it?


The ImageNet data is online: http://www.image-net.org/.

The data descriptions says:

14,197,122 images, 21841 synsets indexed

However, you don't need to train a model on all of the data from scratch, since you can use pre-trained models, e.g. in TF/Keras. See the docs: https://keras.io/applications/.

I don't think the original Alexnet is available, but a lot of other models are ready to load from Keras. You may also find a Alexnet version stored as .hdf5 or so online (which you could load with Keras/TF).

I don't know what you actually want to do, but here is a simple example of how to use a pretrained model (VGG16 in this case) in Keras.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.