Open manthey opened 1 year ago
Currently, we parse annotations by loading the entire annotation into memory. Usinb an iterative approach for larger files (e.g., via https://github.com/ICRAR/ijson) would be mord scalable.
A quick but limited solution to this would be to pick a bigger size than Girder's default of 16MB and read those using the current technique.
Currently, we parse annotations by loading the entire annotation into memory. Usinb an iterative approach for larger files (e.g., via https://github.com/ICRAR/ijson) would be mord scalable.