27 docs tagged with "CCTV"

This document will describe the pipeline of CCTV OCR with reference to the colab notebook.

These are created on the google account hydrotrekgqc@gmail.com

Welcome to CCTV documentation page

The purpose of this document is to track Jake's findings when trying to run LGBM notebooks with new data from DNV.

- We have 43536 images with labels from 73 videos (including Clement Ave video).

- We have 45830 images with labels from 80 videos and 257266 images with labels from 486 videos(including 80 videos)

The following runs are used to evaluate which model configuration perform better. The average F1 and F2 scores of the 10 runs are used as the evaluation metric.

- Link to concept of Fastai

We noticed that a few critical fields, namely Pipe Length, Pipe Size, and Pipe Age,

Validation Scores of Models before fine-tuned with SD1 data

Observation

OCR Libraries

Overview

Here are the important virtual environments for running the CCTV CV and CCTV ML pipelines:

In the gqc-utility-notebooks repo in the MSI machine, there is a gitignored data/ folder. It contains the utilized DNV data and the outputs from the notebooks pertaining to lightGBM.

This document contains Jake's notes of receiving and processing new data from DNV.

- Information on tools used here, like ffmpeg, can be found on general site.

Important links:

02/02/2024

1. How to debug a CCTV script?

11/14/2023

1. Detecting the nearby frames with the defect in view.

There are five videos:

- replace I&I with I&I (water intrusion)

SD1 D dataset

VS_Research folder is synced to local at gqc@192.168.50.57:/media/gqc/External\ HDD which is an external drive, unionsine which is mounted to MicroCenter machine.

We've tested the set of 81 videos and found 5 different video dimensions using this notebook