Description
ATM we are collecting captured videos encoding timestamps in the filename and nearby .log, e.g.
❯ pwd
/home/yoh/proj/repronim/reprostim-reproiner/Videos/2024/10
❯ ls | head
2024.10.01-07.56.48.639--2024.10.01-10.57.51.891.mkv
2024.10.01-07.56.48.639--2024.10.01-10.57.51.891.mkv.duct_info.json
2024.10.01-07.56.48.639--2024.10.01-10.57.51.891.mkv.duct_usage.json@
2024.10.01-07.56.48.639--2024.10.01-10.57.51.891.mkv.log
we should establish the index, where for each video we have a single record with timing information and, if extracted, information about QR codes detected in the video. So pretty much what @vmdocua already has in *qrinfo.jsonl
file for the summary, but then with QR codes embedded inside for the record (strip away video_
prefix which is omni present anyways in this type of record) -- videos_index.jsonl
{
"video_full_path": "Videos/2024/10/2024.10.04-09.36.21.256--2024.10.04-09.41.26.388.mkv",
"video_file_name": "2024.10.04-09.36.21.256--2024.10.04-09.41.26.388.mkv",
"video_isotime_start": "2024-10-04T09:36:21.256000",
"video_isotime_end": "2024-10-04T09:41:26.388000",
"video_duration": 304.85,
"video_frame_width": 1920,
"video_frame_height": 1080,
"video_frame_rate": 60.0,
"video_frame_count": 18291
}
Should be sorted by isotime_start
.
This is purely on information about video. It could be further enhanced with information about QR codes detected in each video but likely we better store it in yet another file -- qrcodes_index.jsonl
where common index would be on the file_name
in the records:
{
"filename": "2024.10.04-09.36.21.256--2024.10.04-09.41.26.388.mkv",
"qr_codes": [ ... ],
"qr_count": 16
}
Similarly we would be establishing index of DICOMs ... TODO: finish up