facebookresearch/hiera
Hiera: A fast, powerful, and simple hierarchical vision transformer.
PythonApache-2.0
Issues
- 1
- 0
COCO pretrained weights
#35 opened by tkhe - 6
Training scripts
#28 opened by VGrondin - 5
Is it possible to replace the backbone from hiera to other hierarchical models like Swin Transformer or CAFormer?
#33 opened by Andy-Ko-0620 - 3
Hiera available on 🤗 Transformers!
#34 opened by EduardoPach - 14
n_output_channels
#32 opened by liushawn618 - 5
- 2
Video pretrained checkpoints with 8 frames?
#24 opened by MLDeS - 0
Pretrained Weights for Something Something V2
#30 opened by edessa - 2
- 1
inference guide for ava.v2.2 dataset
#20 opened by faezehprb - 3
Visualizing MAE predictions?
#17 opened by Wiqzard - 2
- 5
Request for training script
#13 opened by omchakrabarty - 2
Doubt regarding Batch Size
#22 opened by owaisCS - 0
Changing the temporal dimension
#23 opened by MLDeS - 1
how to change number of classes from kinetics to other multiclass dataset?
#21 opened by mbilalshaikh - 8
When will release mae pretrained weights?
#14 opened by shi-lt - 1
- 6
Hiera for Feature extraction
#4 opened by mbilalshaikh - 2
How to save the model ? Since the training code has not been released yet , So I want to train the model myself. I am wondering how should I save the model correctly. Looking forward to your reply!
#16 opened by sulizhi - 2
In the inference.ipynb, each video is extracted to a length of 5s, but how to deal with if the video sizes are different and some videos are short than 5s?
#15 opened by sulizhi - 1
How fine-tune the model?
#12 opened by SalvatoreRa - 4
Getting error with Hiera Inference (Image)
#11 opened by sudhir2016 - 1
- 2
How do we drop tokens?
#5 opened by luke-mcdermott-mi - 1
Example notebook for video task
#7 opened by itsprakhar - 1
comparision with yowo model
#8 opened by ChethanN01 - 0
temporal action recongnition
#9 opened by ChethanN01 - 1
Typo in Readme
#6 opened by luke-mcdermott-mi - 0
- 3
Weird inference time
#1 opened by alexcbb - 0
Add keypoint detection
#2 opened by vincentmin