Datasets:
Sub-tasks:
image-captioning
Multilinguality:
multilingual
Size Categories:
1M<n<10M
Language Creators:
found
Annotations Creators:
machine-generated
ArXiv:
License:
Link to embedding model?
#5
by
dhruv-anand-aintech
- opened
Will it be possible to provide a link to the exact resnet-50 model that was used to extrct the image embeddings? It'll allow people to use this dataset with the their own input images.
Thanks
is it this:
from transformers import ResNetModel
import torch
model = ResNetModel.from_pretrained(“microsoft/resnet-50”)
pixel_values = torch.randn(1, 3, 224, 224)
outputs = model(pixel_values)
features = outputs.pooler_output.squeeze()