Decoupled-Adaptation-for-Cross-Domain-Object-Detection
Decoupled-Adaptation-for-Cross-Domain-Object-Detection copied to clipboard
Questions about split sets of SIM10K dataset
Hello, thanks for your excellent paper and code! I want to follow your great work but got some problems. I download the image set of SIM10K dataset from the official website and found the split is the combination of both the training set and the validation set. I want to know how the authors split the official set into a training set and a validation set?
Sorry for the late reply. To be consistent with previous work, trainval10k is used both in training and validation (Note that no annotations of Sim10k are used during training).