search

tnc-ca-geo / animl-ml

Machine Learning resources for camera trap data processing
Other
4 stars 1 forks source link

Megadetector endpoint can return objects outside of images' dimensions #103

Closed nathanielrindlaub closed 1 year ago

nathanielrindlaub commented 1 year ago

e.g. this image returns:

[
  {
    x1: -0.002311658812686801,
    y1: -0.021098900586366653,
    x2: 1.0007762908935547,
    y2: 1.0159189701080322,
    confidence: 0.3330053687095642,
    class: 2
  }
]

@rbavery thinks this should soon be addressed by the letterbox fix, but we should double check once that is live.

rbavery commented 1 year ago

I think this is addressed after we fixed the shifted bbox issue.