Hey, Thanks a lot for putting this dataset together and congrats on getting accepted into the NeurIPS workshop. I'm going to be poking around the repo, and I'll post issues as I see bugs. I'm sure you're busy getting everything together for camera ready, so don't worry about debugging these in real time. Here's the first round that I found when trying to run download_datasets.py:
First, the PovertyMapDataset class has a typo on the declaraion of the versions_dict variable, line 124. The download_urls entry should contain a list, but instead contains a dictionary without keys.
Second, the initialize_data_dir() function in the SustainBenchDataset class requires versions_dict to contain two keys which are both mission from the PovertyMapDataset: download_url and compressed_size.
Third, self._data_dir is assigned incorrectly in crop_seg_dataset.py.
Hey, Thanks a lot for putting this dataset together and congrats on getting accepted into the NeurIPS workshop. I'm going to be poking around the repo, and I'll post issues as I see bugs. I'm sure you're busy getting everything together for camera ready, so don't worry about debugging these in real time. Here's the first round that I found when trying to run download_datasets.py:
First, the PovertyMapDataset class has a typo on the declaraion of the versions_dict variable, line 124. The
download_urlsentry should contain a list, but instead contains a dictionary without keys.Second, the
initialize_data_dir()function in the SustainBenchDataset class requires versions_dict to contain two keys which are both mission from the PovertyMapDataset:download_urlandcompressed_size.Third, self._data_dir is assigned incorrectly in crop_seg_dataset.py.