Size of saved model too big (~ 2 gb)

microsoft / multiview-human-pose-estimation-pytorch

This is an official Pytorch implementation of "Cross View Fusion for 3D Human Pose Estimation, ICCV 2019".

MIT License

549 stars 89 forks source link

Size of saved model too big (~ 2 gb) #27

Closed rlrahulkanojia closed 4 years ago

rlrahulkanojia commented 4 years ago

the final model saved on a mobile net backbone is of 2 GB. Can someone. please help me where I might be wrong. I have not made any significant changes in the code.

CHUNYUWANG commented 4 years ago

The model is big because of the fusion layer which is implemented by fully connected layer.

rlrahulkanojia commented 4 years ago

Yes, I found it after a detailed analysis. But the role of optional aggregations layers is not clear to me. It would be very helpful if you could provide any clarification.

CHUNYUWANG commented 4 years ago

We will fuse features across different views if the optional aggregation parameter is true. You can take a look at the code [lib/models/multiview_pose_resnet.py]