voc_eval.py on my own dataset

[Kasyene]

Hello,
I introduced my own dataset with 1 real class + background class.
My dataset is in voc-like format.
I trained it, I can test it and it detects trained class without any problems.
The eval_ssd.py generates the results file. The problem I have is with voc_eval.py to measure map for my class.

VOC07 metric? Yes
Traceback (most recent call last):
  File "voc_eval.py", line 268, in <module>
    do_python_eval()
  File "voc_eval.py", line 79, in do_python_eval
    ovthresh=0.5, use_07_metric=use_07_metric)
  File "voc_eval.py", line 221, in voc_eval
    R = class_recs[image_ids[d]]
KeyError: '0594'

I am not sure if I understand what test.txt should contain and how to fix this issue.
I attatched files with my dataset and result.txt I got from eval_ssd. test.txt attatched is the original one - with 1st error.

trainval.txt
val.txt
test.txt
train.txt
results_1.txt

Edit: I assumed the probem was that file 0594 was in result file and not in the test.txt. So I tried replacing file names in test.txt with all the names in results.txt (without duplicates) and it works now. I am still unsure because I was sure test.txt should contain different set of files.

Can you please help? It seems I am misunderstading something completely.