question on the model settings in Tab. 5

[twangnh]

Hi Xingyi, thanks for your wonderful work! I have a question regarding the model setting, in Tab. 5, for the naive merge, how is the merge conducted? and for the retrained, in my understanding, the box labels of multiple datasets are converted to the unified space and used for training a unified model, but still, the objects in one dataset may not be labeled in another, so we could still face such background confusion in supervision? is my understanding correct?

Thanks in advance!

[xizaoqu]

Hi Xingyi, thanks for your wonderful work! I have a question regarding the model setting, in Tab. 5, for the naive merge, how is the merge conducted? and for the retrained, in my understanding, the box labels of multiple datasets are converted to the unified space and used for training a unified model, but still, the objects in one dataset may not be labeled in another, so we could still face such background confusion in supervision? is my understanding correct?

Thanks in advance!

Hi! I am also confused about this part. Have you figured it out? In my understanding, naive merge stands for merging the predictions from partitioned heads to a unified label space. Thus the results will inevitably worse than directly predicting to a specific datasets. retrained may be mean "For each label space, we retrain a multidataset detector with that label space" as the author said in the paper. However, doesn't this introduce priors on the corresponding dataset?