Fault Sense

An image processing/analysis application. for the purposes of fault detection of products produced in real time environments. Designed to be deployed in isolation on an embedded device.

---

Dataset Overview

This project is developed specifically for use on the VisA (Visual Anomoly Dataset).

Object	# normal samples	# anomaly samples	# anomaly classes	object type
PCB1	1,004	100	4	Complex structure
PCB2	1,001	100	4	Complex structure
PCB3	1,006	100	4	Complex structure
PCB4	1,005	100	7	Complex structure
Capsules	602	100	5	Multiple instances
Candle	1,000	100	8	Multiple instances
Macaroni1	1,000	100	7	Multiple instances
Macaroni2	1,000	100	7	Multiple instances
Cashew	500	100	9	Single instance
Chewing gum	503	100	6	Single instance
Fryum	500	100	8	Single instance
Pipe fryum	500	100	9	Single instance

Attribute	Total
Objects	12
Samples	10,821
Normal Samples	9,621
Anomoly Samples	1200
Unique Anomoly Classes

A more comprehensive breakdown and sample images from specific classes can be obtained in the following document.

---

Building & Running the Project

Data Preparation

From the project root execute the following commands in order.

cd build/

cmake .

make

./data-preperation

Building CLI Interface

From the root of the project run.

cd src/main/frontend/build/

cmake .

make

Running Application

the fault-sense-cli executable will then be available in the current working directory.

To view available commands execute:

./fault-sense-cli --help

Example command:

./fault-sense-cli view -i ../../../../data/sample-images/chewinggum-anomoly.JPG

Note: press 'q' to exit

---

Dependencies

1. opencv2
2. Catch2 (for testing)

---

Further Information

Name	Description
dataset-examples-and-descriptions	Contains a detailed breakdown of the dataset. Including overview statistics and object and class breakdowns.
project-managment	Contains links to the tools used for tracking and planning project progression
testing	Contains descriptions on how and where to build, run and add to the project test suite.