handong1587

Does anyone get an optimal result on KITTI?

4

Cannot find any possible results on KITTI, I am wondering if someone already reproduce the results of original paper.

I can only obtain top-1/mAP = 68.62%/68.13% performance. Anything wrong?

3

Detailed evaluation result: ``` all detection: recall = 76.39% ap = 70.73% labeled only detection: recall = 96.74% search ranking: mAP = 68.13% top- 1 = 68.62% top- 5 =...

In `utils.py`, I guess you should put line151 before line143 to get `annotated_attributes` correctly? [https://github.com/bethgelab/openimages2coco/blob/master/utils.py#L151](https://github.com/bethgelab/openimages2coco/blob/master/utils.py#L151) Since after line143, `original_annotations` will pop out original_annotations[0] then you lose that 5 attributes.