Fail to evaluate

[cocoshe]

Here's the error when evaluating

in _generate_matrix
    confusion_matrix = count.reshape(self.num_class, self.num_class)
ValueError: cannot reshape array of size 15377 into shape (124,124)

[cocoshe]

And could you tell me where is the Spatial-Temporal Fusion part since I failed to find the code?

[bingnoi]

please give me more detail, so i can help you solve the problem.

[cocoshe]

please give me more detail, so i can help you solve the problem.

Thx for your reply!
I just want to ask about this part in paper, how can I find the corresponding code?

[cocoshe]

And how you use SAM in the setting?

[bingnoi]

1.the code is basically refers to https://github.com/AVC2-UESTC/OV2VSS/blob/master/mmseg/models/decode_heads/hypercorre.py
2.SAM does not support open-text prompts. We followed the common practice of first using SAM to segment all relevant regions, and then used CLIP for classification

[cocoshe]

1.the code is basically refers to master/mmseg/models/decode_heads/hypercorre.py 2.SAM does not support open-text prompts. We followed the common practice of first using SAM to segment all relevant regions, and then used CLIP for classification

Tkx for your reply, and another question is ViT-B2, could you tell me about it(where can I find it), it seems not commonly used checkpoint? Or it actually some kind of typo here?