Sam1 is giving better results in comparision to Sam2

[plutus123]

So I have had used sam1 and sam2 for image segmentation and compared their results.
The image segmentation done by sam1 is much better than sam2.

Result of sam1:

print(len(masks))

Output:

16

Result of sam2:

print(len(masks))

Output:

1

Please can anyone tell me why there is so much difference in the results from both the models. Also suggest me any possible points that I must take into consideration in order to improve the performance of my sam2 model.

[amirmohammadnsh]

I think you can follow up your question in this issue #93 with the similar topic.

[plutus123]

@amirmohammadnsh Sam2 is unable to detect correct number of masks like in above example that I have had provided sam1 was predicting 16 masks whereas sam2 was detecting only 1 mask.

[thedatamask]

Yes, check out this video for more details. Very useful where they have compared SAM 1 vs SAM 2 with demo.

SAM vs SAM 2.0: Meta's Segmentation Models . When to use what ? Case Study

https://www.youtube.com/watch?v=vMI-TnyNLYU

Please do like and subscribe to the channel.