Skip to content
This repository has been archived by the owner on Apr 19, 2021. It is now read-only.

Model Optimization #8

Open
Waidhoferj opened this issue Feb 2, 2021 · 0 comments
Open

Model Optimization #8

Waidhoferj opened this issue Feb 2, 2021 · 0 comments
Assignees
@ShainaBagri @dahlgreb
Labels
API 🤝 Involves communications between JavaScript, Python and the Argo server Machine Learning 🦾 Involves the architecture of computer vision models Research Look into this issue, weigh alternatives and report back in the comments.
Milestone
Depth Estimation

Comments

@Waidhoferj
Copy link
Member

Can we run models on a local instance of Argo Editor? We have been running into performance issues for MiDaS. Ideally, the model should do its thing in 5-7 seconds max.

Areas of research:

  • Can we get TF to work with a GPU (maybe tensorflow-gpu)
  • Is there a smaller/more performant version of the MiDaS model? We're currently using this one
  • If we can't run it on the frontend, should we mount our model in a cloud instance where GPU resources are avaliable? What are our options?
@Waidhoferj Waidhoferj added Machine Learning 🦾 Involves the architecture of computer vision models API 🤝 Involves communications between JavaScript, Python and the Argo server Research Look into this issue, weigh alternatives and report back in the comments. labels Feb 2, 2021
@Waidhoferj Waidhoferj added this to the Depth Estimation milestone Feb 2, 2021
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees

@ShainaBagri ShainaBagri

@dahlgreb dahlgreb

Labels
API 🤝 Involves communications between JavaScript, Python and the Argo server Machine Learning 🦾 Involves the architecture of computer vision models Research Look into this issue, weigh alternatives and report back in the comments.
Projects
None yet
Milestone
Depth Estimation
Development

No branches or pull requests
3 participants
@ShainaBagri @dahlgreb @Waidhoferj