/SigmoidFocalLoss_cuda.cu error about alphaction HOT 5 CLOSED

This issue seems to be a duplicate of issue #28 . Please try the suggestions provided in the link #28 (comment).

Thank you for quick reply! It helps, I updated the pytorch to version 1.3.0, the error disappeared.
But there is a new error. when I run demo.py using the pretrained model common_15cat_res101.pth, the computer shut down as it processed to the 33th frame. I try to find the reason,but can't figure out why it coused the computer to reboot. Looking forward for your reply!
the command I used is as follow:
python demo.py --video-path Fighting033_x264.mp4 --output-path ./output.mp4 --cfg-path ~/ZSJ/action_detection/AlphAction/config_files/resnet101_8x8f_denseserial.yaml --weight-path ~/ZSJ/action_detection/AlphAction/data/models/aia_models/common_15cat_res101.pth --common-cate

Computer shutting down is due to the power protection of GPU. I sovled this problem by set the limitation of GPU to 150w by using the command:
sudo nvidia-smi -pl 150

from alphaction.

This issue seems to be a duplicate of issue #28 . Please try the suggestions provided in the link #28 (comment).

from alphaction.

Mybe I can give u some help , you can try conda create -n alphaction python=3.8, Don't use python3.7.

from alphaction.

This issue seems to be a duplicate of issue #28 . Please try the suggestions provided in the link #28 (comment).

Thank you for quick reply! It helps, I updated the pytorch to version 1.3.0, the error disappeared.
But there is a new error. when I run demo.py using the pretrained model common_15cat_res101.pth, the computer shut down as it processed to the 33th frame. I try to find the reason,but can't figure out why it coused the computer to reboot. Looking forward for your reply!
the command I used is as follow:
python demo.py --video-path Fighting033_x264.mp4 --output-path ./output.mp4 --cfg-path ~/ZSJ/action_detection/AlphAction/config_files/resnet101_8x8f_denseserial.yaml --weight-path ~/ZSJ/action_detection/AlphAction/data/models/aia_models/common_15cat_res101.pth --common-cate

from alphaction.

Mybe I can give u some help , you can try conda create -n alphaction python=3.8, Don't use python3.7.

Thank your for suggestion. The main problem is the version pytorch 1.0.0 is not match to the code. Updating the pytorch works~

from alphaction.