Dear organizers,
Would you provide a docker template for submission ? I also would like to know the specification of the inputs and the outputs.
Created by SKJP Hi,
I wanted to give a short update regarding the dockerfile entry point. the entrypoint given ist just a placeholder and is not used/is overridden by our docker run command.
I believe I understand that the confusion regarding this was coming from the main.py file in the evaluation folder. This file is called _after_ the dockers are run to evaluate the metrics on the predictions output by the dockers. To avoid any further confusion with this, I have renamed this file in the evaluation folder to "evaluate_metrics.py".
I hope this resolves any remaining confusion regarding this. Thank you for making us aware of this discrepancy! No problem :)
Thanks for making me aware of the discrepancy between the two instructions. I'll fix this in the wiki.
If you have any other questions or comments, feel free to let me know. Thank you for you clarification. I think I understood now.
I think the instructions in wiki and pdf are different a bit and this causes the confusion. I think other participants may be confused.
Hi,
How you write the shell script is completely up to you. We provide the script name (Process_SkillEval.sh), so we can uniformly call and run the docker containers. The requirements for the script are specified in the [detailed instructions](https://caruscloud.uniklinikum-dresden.de/index.php/s/Tq33KRs54ogWcPM) in section 2.1. How you process the input parameters and what you include in the script is entirely up to you.
Regarding the entry point: we override this when we run the dockers by calling the shell script (Process_SkillEval.sh). To see the commands we call to run the dockers, see section 2.1 from the detailed instructions or the [docker submission page](https://www.synapse.org/Synapse:syn54123724/wiki/626584).
I apologize for the confusion this may have caused. Does this help you further? Thank you for your answer.
Though It's a different topic, I think you don't mention the details of "Process_SkillEval.sh" in the instruction.
I think the shell script would be as follows, should be included in the docker image, and the entry point should not be included in the Dockerfile.
---
\#!/bin/sh
python main.py \$1 \$2
---
Please let me know if my understanding is correct.
Hi,
1. yes, the test directory will have all test videos in one directory.
2. you can call them predictions_GRS.csv (for task 1) and predictions_OSATS.csv (for task 2)
3. in the output csv file the VIDEO field should just be the ID: so if the filename is "001.mp4", the field in the VIDEO column should be "001"
Let me know if you have any further questions Thank you very much. Now I have additional questions.
1. There are multiple videos in an input directory. Right ?
2. What should be the names of the output csv files for both tasks ?
3. In the output csv file, how should we specify the "VIDEO" field ? For example, when the video file name is "001.mp4", is this field "001" or "001.mp4" ? Hi,
General instructions for the docker and submission can be found [here](https://www.synapse.org/Synapse:syn54123724/wiki/626584).
From there you can also find more detailed instructions (including an example Dockerfile for our challenge in section _2.2 Example Docker from Dockerfile_) or also click [here](https://caruscloud.uniklinikum-dresden.de/index.php/s/Tq33KRs54ogWcPM).
A base image can be found [here](https://gitlab.com/nct_tso_public/challenge-docker/-/blob/main/BaseImage/Dockerfile).
I hope this clarifies what you need to know. If you have any further questions, don't hesitate to ask :)