Problems with Evaluations
jmiemirza opened this issue · 1 comments
jmiemirza commented
Hi,
I have been using a new dataset but it has been built on the same format as KITTI. I am having trouble with the evaluation script. When I run the evaluation script to evaluate all the checkpoints, nothing is written in the results file. I am attaching a screenshot:
This is how it looks. However, I tried with just 50 samples of the same dataset and it provided me with the following output file:
Now when I follow the readme and evaluate individual checkpoints, I get only the Car Detection AP and then a segmentation fault.
Did anyone have the same problem?
Thanks
asharakeh commented
@jmiemirza this might be caused by multiple issues. I would recommend the following:
- Check if your results are written appropriatly.
- Check if you have a score value per result at the 16th column and that the score is between 0-1.
- Check if your bounding boxes make sense. 2D bounding boxes should never be negative, 3D bounding boxes should have positive z coordinate etc...