parser & checker (images to video conversion): parsing the json to retrieve information about tiff images

[madvid]

User Story

I want to convert any .TIFF image to AVI and MP4 types video. so that i can use these data for WF NTP

To do so, I need to parse an argument being supposely the path to the directory containing:

• several tiff images
• 1 json which may be metadata concerning the video settings.

This part is dealed in the issue #17 and is the first step realised by the execution of:

python -m converter --path [path_arg]

Based on path_arg I want that the program converter.py to parse the json file and retrieve the different information to check the validity of the tiff files.

The handle of the variable frame rate is tricky (related issue: #24 ). Thus to level of development is expected:

• average the frame rate a construct a video with this average.
• use the instant frame rate

It may be important for futur analysis to keep a track of the instant frame rate. Two choices are possible:

• generate a json with the time interval between consecutive frame
• develop the functions to easily retrieve this information from the json file containing the metadata video. This choice is at the developper discretion.

Acceptance Criteria

• based on the json file check if
  • [x] .tiff files are valid (may need to check if there is header in tiff file and check it).
  • [x] .tiff files are of the same size
  • [x] .tiff files have the same pattern (or the name is identical to those one can find in the json)
  • [x] there is no missing tiff
  • [ ] check minimum number of files to convert (100 files)
  • [ ] retrieve frame rate from the json file.

Definition of Done

• [ ] Product Owner accepts the User Story
• [ ] Acceptance criteria met

Defined by CICD:

• Code reviewed
• black compliant
• Unit tests passed
• Functional tests passed
• Non-Functional requirements met

[COTHSC]

I have an good start for all of this except the frame rate. The time between pictures can vary by ~20 ms meaning the frame rate is variable. It is easy to collect exact timestamps for each picture, however writing a variable frame rate video may be a bit trickier (but definitely possible), and the tracking algorithm would need to take the variable frame rate into account. Otherwise we could average out & approximate frame rate.

[madvid]

Nice ! I will update the issue a bit to segment the work into several level of complexity.

[madvid]

@COTHSC user story is updated

[madvid]

@COTHSC and @rcatini, for the tests you can perform some of them to start with the URI:

• s3://lab-nematode/tests/test_1/ is valid
• s3://lab-nematode/tests/test_2/ is invalid because name of a tiff is modified
• s3://lab-nematode/tests/test_3/ is invalid because a file tiff has been deleted
• s3://lab-nematode/tests/test_4/ invalid because the size of an image has been modify in the metadata.json

[p-lg-ux]

Instead of a list of the time intervals between the frames, it could be better to have a dictionary with the format dict = { "filename0.tiff": interval0_ms, "filename1.tiff": interval1_ms, ..., "filenameN.tiff": intervalN_ms } where interval0_ms is equal to 0.

[cnstll]

@COTHSC @rcatini Or we could have the following dict in output from your parsing functions:

dict = {
'frame_names': ['filename0.tiff', 'filename1.tiff', ..., 'filename100.tiff'],
'frame_height': 2048
'frame_length': 2048
'avg_fps': 10
}