Detection competition - unable to submit predictions to CodaLab

[Britefury]

Hi,

I've been having difficult submitting to the detection competition. It appears that my text file format is incorrect, as I get the following error:

Traceback (most recent call last):
  File "/tmp/codalab/tmpxfH0zT/run/program/evaluate.py", line 438, in 
    detection_evaluation(truth_file, source_file, adaptation_file, output_file)
  File "/tmp/codalab/tmpxfH0zT/run/program/evaluate.py", line 363, in detection_evaluation
    for line in f.read().splitlines()
  File "/tmp/codalab/tmpxfH0zT/run/program/evaluate.py", line 21, in chunks
    raise RuntimeError("Sequence does not split into {k}, because last elem has {elem.index(uniq)} left")
RuntimeError: Sequence does not split into {k}, because last elem has {elem.index(uniq)} left

Would it be possible for you guys to either:

• provide an example submission file that the server can handle
• provide code to generate a valid submission file
• provide the source code of 'evaluate.py' that runs on the server
• modify the evaluation code so that it reports more helpful error messages

Thanks! :)

[neelakaushik]

Hi, Sorry you have had trouble submitting! We have uploaded the evaluation code here as well as the ground truth file for validation here and a sample predictions file here. If you are still having trouble submitting, please feel free to share your submission file with us and maybe we can try to figure out the issue from there.

[Britefury]

Thanks! Figured out my mistake.

[MInner]

Just to clarify, *.pkl file is just a "precomputed cached" bbox data inferred from gt files and images provided previously. It should be portable, but I am not 100% sure. Code in main github repo generates it automatically from gt files.

[Britefury]

Referring to the other issue (https://github.com/VisionLearningGroup/visda-2018-public/issues/10) the pkl file loads fine using the pickle module. The bbox values in the pkl file are scaled when compared to those in coco17-val.txt.