I downloaded the Mapillary Traffic Sign Dataset. The images as well as the annotations have the “old” image_key, which was deprectaded in v4. I wanted to search more metadata for each image of the dataset using the API, but it’s not possible, since the available image_key is not supported by the API, which supports image_id. A user proposed to access the current image_id by sending a request with the old image_key, but I do not want to send 40.000 requests.
Is there a complete list to link the old image_key with the new image_id?
Cool, it looks like about ~3,300 out of the ~24,000 are missing - I’m not sure why that is - I think you could do the requests to get them, but maybe having 20k gives you enough to work with already?
The table contains 200,538 entries. I extracted the file names of the annotation JSONs (under mtsd_fully_annotated_annotation) and obtained 41,909 image keys. I then performed an inner join in R to find matches between the table and the dataset’s image keys, but found only 19 matches.
This means that 41,890 image IDs are missing, which I could potentially scrape. However, I am unsure whether sending 40,000 requests within an hour might cause any issues.
If this volume of requests is acceptable, I can provide you with an updated table containing complete image_key and image_id values.