Open opencvscikit opened 10 months ago
The training and test(validation) dataset are generated from random samples of the global map and used separately in the model training. The global map is also processed using the final trained model, since we want to map all craters on Mars.
I looked at the author's Github code and found that your test dataset seems to be a dataset obtained from DEM global segmentation. But in this way, your training dataset includes the test dataset. Is it because I misunderstood the code.