torchvision==0.11.0 is yanked help!!!

[TanishqAgarwal29]

Hello Owner,

When I am using the code:

!conda install mamba -c conda-forge # install mamba in base environment
!mamba create -n hoi_torch110 python=3.9 -c conda-forge
!conda activate hoi_torch110
!mamba install pytorch==1.10.0 torchvision==0.11.0 torchaudio==0.10.0 torchtext==0.11.0 cudatoolkit=11.3 matplotlib numpy opencv Pillow PyYAML requests scipy tqdm tensorboard wandb pandas seaborn Cython easydict gdown scikit-learn imageio notebook ipywidgets black -c pytorch -c conda-forge

Its not getting executed as torchvision==0.11.0 is yanked means its not available so there is no use of torch version 1.10.0 so its not getting executed in google colab, please suggest me a different code for this.

Warm Regards,
Tanishq Agarwal

[nizhf]

Hello,

Thank you for your interest in our work!
I tried this code on my local machine. It seems like opencv causes the incompatibility problem.
Maybe you can use pip instead of mamba to install opencv.
Simply remove opencv from that mamba install ... line, then execute pip install opencv-python.
I tested some cv2 image operations and they worked properly. Please let me know whether this also works for your environment.

Best,
Zhifan Ni

[TanishqAgarwal29]

I am getting a new error when I am executing the next line of code

Detection

%cd /content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer
!python run.py --source /content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/ABA Therapy_ Daniel - Communication.mp4 --out /content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/rohan --future 1 --hoi-thres 0.3 --print

For anticipation, set future to 1, 3, 5, or 7

AND THE ERROR I AM GETTING IN GOOGLE COLAB IS

Detection

%cd /content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer
!python run.py --source /content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/ABA Therapy_ Daniel - Communication.mp4 --out /content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/rohan --future 1 --hoi-thres 0.3 --print

For anticipation, set future to 1, 3, 5, or 7

/content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer
Traceback (most recent call last):
File "/content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/run.py", line 19, in
from common.transforms import STTranTransform, YOLOv5Transform
File "/content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/common/transforms.py", line 7, in
from modules.object_tracking.yolov5.utils.augmentations import letterbox
File "/content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/modules/object_tracking/init.py", line 4, in
from .feature_backbone import FeatureExtractionResNet101, FeatureExtractionResNet50
File "/content/drive/MyDrive/Colab Notebooks/hoi-prediction-gaze-transformer/modules/object_tracking/feature_backbone.py", line 7, in
from torchvision.models.resnet import ResNet, Bottleneck, model_urls
ImportError: cannot import name 'model_urls' from 'torchvision.models.resnet' (/usr/local/lib/python3.10/dist-packages/torchvision/models/resnet.py)

PLEASE SUGGEST ME I NEED THIS MODEL TO WORK I HAVE SOME REALLY URGENT NEED OF THIS MODEL PLEASE HELP ME OWNER!!

[nizhf]

Which torchvision version are you using? From maybe 0.14.0, pytorch removes that model_url.