search

yoxu515 / aot-benchmark

An efficient modular implementation of Associating Objects with Transformers for Video Object Segmentation in PyTorch
BSD 3-Clause "New" or "Revised" License
600 stars 108 forks source link

To feed truth values into the model during the testing phase? #81

Open lbbcalm opened 2 months ago

lbbcalm commented 2 months ago

Hello, this job is awesome, I would like to ask, does this model need to feed truth values into the model during the testing phase?

z-x-yang commented 2 months ago

What's the meaning of "feed truth values"?

lbbcalm commented 2 months ago

Thanks for the reply, it means that for training, you need to feed the training image and the truth of the first frame into the model, then for testing do you still feed the test set and the truth of the first frame or do you just need to feed the test image to get the prediction map?

z-x-yang commented 2 months ago

The first-frame mask is required to tell the model what's the target/object during inference.

lbbcalm commented 2 months ago

Okay, thanks.