TimeSformer (base-sized model, fine-tuned on Kinetics-600)
TimeSformer model pre-trained on Kinetics-600. It was introduced in the paper TimeSformer: Is Space-Time Attention All You Need for Video Understanding? by Tong et al. and first released in this repository.
Disclaimer: The team releasing TimeSformer did not write a model card for this model so this model card has been written by fcakyon.
Intended uses & limitations
You can use the raw model for video classification into one of the 600 possible Kinetics-600 labels.
How to use
Here is how to use this model to classify a video:
from transformers import AutoImageProcessor, TimesformerForVideoClassification
import numpy as np
import torch
video = list(np.random.randn(16, 3, 448, 448))
processor = AutoImageProcessor.from_pretrained("facebook/timesformer-hr-finetuned-k600")
model = TimesformerForVideoClassification.from_pretrained("facebook/timesformer-hr-finetuned-k600")
inputs = processor(images=video, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
logits = outputs.logits
predicted_class_idx = logits.argmax(-1).item()
print("Predicted class:", model.config.id2label[predicted_class_idx])
For more code examples, we refer to the documentation.
BibTeX entry and citation info
@inproceedings{bertasius2021space,
title={Is Space-Time Attention All You Need for Video Understanding?},
author={Bertasius, Gedas and Wang, Heng and Torresani, Lorenzo},
booktitle={International Conference on Machine Learning},
pages={813--824},
year={2021},
organization={PMLR}
}
- Downloads last month
- 2,502