Current Status of DNV Utility

DNV file locations

No Current Database

There is no current database that I would be able to use for the (720?)

What tables will i need to Run Compute MSI predict. I do not need to run anything before this as I don't need to create the model because I will be using an already generated model.

Look at W&B for DNV can I figure out the best model from there? Accurascy multi is the best thing to look at

I think I have most of the videos table. from the CSV I will need the frames table. Look into the OCR JSON files/ blurred and unblurred images for the video groups if they are not in the Videos CSV.

792

This is all the DNV data. It is comprised of all the other groups. 792 = 270 + 35 + 406 + 81 DNV_A (dnv_pacp_A_81) DNV_B (dnv_pacp_B_406 ,) 270 PACP Videos (PACP_270) 35 PACP Videos (dnv_pacp_35) https://projects.gqc.com/internal/CCTV/CCTV%20Pre-2024/Data/dataset_summary

The Process for DNV.

We want to run a model that already exist on DNV 792. To do that I will need to create a database that can run predict on 792. Some tables can be copied from Other tables like Conditions Standard

I will need to construct the files for 792.

1. Blurred Frames zip string. They should be sorted by video group.
2. Azure JSON File should be stored by same video groups in zip files.
3. Streamlit needs a copy of all frames unzipped and unblurred for Carcerosel.
4. File organization on Data0 will be to have a DNV folder.
   1. Create an MD file and put it into this folder explain what is in there and why explain how we are getting 792 and any other documentation that needs to be in this folder. It should be know that if we make runs for PACP in DNV then this is where we want to reffrence the files.
   2. Blurred Frames.zip
      1. 406 -/media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/Blurred_Frames/
      2. 35 - /media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/Blurred_Frames/
      3. 270 - /media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/PACP_270/Data/Blurred_Frames/
      4. 81 - /media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/CCTV_Training_Data/
   3. Azure Json.zip
      1. 406 -/media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/Azure_JSON/
      2. 35 - /media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/Azure_JSON/
      3. 270 - /media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/PACP_270/Data/Azure_JSON/
      4. 81 - /media/gqc/unionsine1/VS_Research/CCTV/DNV/Data/Azure_JSON/
   4. Unzipped un-blurred frames.

What Tables are required to run prediction after a model is generated.

• videos (Streamlit)
• frames (Streamlit)
• Models (Streamlit)
• All conditions
• Video Groups (Streamlit)
• Video Types?
• Regex Distance
• Distance Rectangle
• Condition Code Rectangle
• Condition Standards. (Streamlit)
• Check predict to make sure I don't need anything else.

What notebooks do I need to run.

From COV YAML

  VIDEO_LIST_CSV_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/Video_Lists/{video_group}.csv'
    EXTRACTED_FRAMES_ZIP_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/blurred_frames/{video_group}.zip'
    BLURRED_FRAMES_ZIP_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/blurred_frames/{video_group}.zip'

    AZURE_JSON_ZIP_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/Azure_JSON/JSON_COV_{video_group}.zip'

    LABELS_CSV_ZIP_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/Labels_CSV/extracted_data_{video_group}.zip'
    VIDEO_DB_PATH_STRING: '{WORKING_PATH}/{UTILITY}/Video_DB/Video_DB_COV.db'
    RECEIVED_VIDEO_DIR_STRING: '{WORKING_PATH}/{UTILITY}/{DATASET}/Data/received_data/'
    LOG_DIR_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/Logs/'
    TEMP_DIR_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/temp/'
    TRAINING_DATA_CSV_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/training_csv/train_{UTILITY}_{DATASET}_{SPLIT_BASIS}.csv'
    TEST_DATA_CSV_STRING: '{WORKING_PATH}/{UTILITY}/Data/{DATASET}/training_csv/test_{UTILITY}_{DATASET}_{SPLIT_BASIS}.csv'