search

SHI-Labs / Matting-Anything

Matting Anything Model (MAM), an efficient and versatile framework for estimating the alpha matte of any instance in an image with flexible and interactive visual or linguistic user prompt guidance.
https://arxiv.org/abs/2306.05399
MIT License
615 stars 49 forks source link

How long it takes to train MAM? #5

Closed KainingYing closed 1 year ago

chrisjuniorli commented 1 year ago

As mentioned in the paper, we trained MAM with SAM ViT-B checkpoints frozen on 8 RTX A6000 GPUs, with a batch size of 10 images per GPU. The total 20,000 iterations take ~15h.

KainingYing commented 1 year ago

Thank you for your generous reply.