Few questions for adaptation strategy.

[BOBrown]

@yuhuayc For domain adaptation of detectors from the source domain to the target domain, we always train from the ImageNet pre-trained model. However, in practical application, the pre-trained model on the source domain is usually available. Why don't we fine-tune from the pre-trained model on the source domain model, but fine-tune from the ImageNet pre-trained model. The latter seems to take more time. Could you explain the reason for this?

[heekhero]

@BOBrown I think the reason why we choose to load the model pre-trained on ImageNet rather dataset-specific model is just for simplification. You can get enough semantic information from the existing ImageNet and that's what we need all. Of course, I think, maybe a dataset-specific model does result in a faster training process, but that should not be the research focus.