mAP is 0 for all the classes detected

[arthur-telles3]

Hi Rafael,

I am pretty sure I followed all the instructions properly but still I can't get another result rather then 0 mAP for all my classes. I am attaching the detection and groundtruth files and the csvs that I used to generate the txt files.

I issued the following command as my boxes are in the configuration

python pascalvoc.py -gt groundtruths_blindenhund_test -det detections_blindenhund_test -gtformat xyrb -detformat xyrb

Really appreciate the help given.

csvs with the boxes.zip
detections_blindenhund_test.zip
groundtruths_blindenhund_test.zip

[rafaelpadilla]

Dear @Art31 ,

In your filtered_groundtruth_test.csv I notice something strange. The first row, for instance has: bottom = 330, left = 121, top = 584, right = 768

I'm reading it as: x1 = 121, y1 = 584, x2 = 768, y2 = 330

How can the bottom coordinate (y2) be less than the top coordinate (y1)? It happens in many (if not all) your bounding boxes.

Please, check:

1. You should take the origin of your image (0,0) as being the upper left corner.
2. As you are using the parameters -gtformat xybr and -detformat xyrb, your text files must be in the format: <filename> <confidence> <left> <top> <right> <bottom>

Please, close this issue if it solves your problem.

[arthur-telles3]

You are right, it was my mistake, my apologies. Thank you for the help!