How To Submit - Konnsy/REAML2022-hackathon GitHub Wiki

For all submissions:

  • Submit the code that can be used directly, i.e. test code. We will not train your models but execute it in the evaluation mode!
  • Submit all needed files. Do not forget to include the trained model weights (if you trained anything).
  • A submission must be packed into in a .zip file (no loose files).
  • Give the files that you want to submit unique names (name of your team or team members and time of submission).
  • You may submit multiple times, but make sure to specify the new submission time.
  • We will take the score of your latest submission.

Pay attention to the time limits

  • Finish the first task before 17:00
  • Finish the second task before 20:00
  • Finish the third task before 22:00

Submission for Task 1:

  • Your system has to predict a confidence score (<0.5: only background, >0.5: particle) for each given block of raw images.
  • Submit your final version of Task 1 before working on Task 2. This makes it easier to point out problems with the submission at an early stage.

Submission for Task 2:

  • Your system has to predict a number of particles that it detects in a whole dataset given by its folder path.

Submission for Task 3:

  • Same as in Task 2 with the restriction to be executable on one of the Odroid devices
  • If you want to be safe: show us the used device and we copy its SD card to have an exact copy of the system

Submission formats

Submission format of "test_results.txt" for task 1

Please submit a file where each line contains an entry <final result (bool)>, <positive cases (int)>, <negative cases(int)>,

An example would be
"False, 23, 123, test_set_B
True, 45, 2, test_set_F"
for two test sets "test_set_B" and "test_set_F" where the "final result"/first column is True if and only if there are more positive than negative cases.

Submission format of "test_results.txt" for task 2

Please submit a file where each line contains an entry <number of particles (int)>, , <time to compute in seconds (float)>

An example would be
"123, test_set_C, 78.4
0, test_set_E, 87.5"
for two test sets " test_set_C" and " test_set_E"

Try to keep your models explainable so that a plausibility check is possible for your submission.