Pipeline training and inference on UI

[vnk8071]

Pipeline Anomaly Detection in Industry Manufacturing on UI

Overview

Anomalib is a deep learning library that aims to collect state-of-the-art anomaly detection algorithms for benchmarking on both public and private datasets. With my pull requests #729 and thank @ashwinvaidya17 and @jpcbertoldo for suggestion.

In this project, I create a pipeline for training and inference with all datasets format MVTec (custom) on UI. Everyone can training with all Anomalib models supported and quickly to demo. And my project is an end-to-end example in anomaly detection task.

For more details, everyone can visit and use my project below link.
Link my project: https://github.com/vnk8071/anomaly-detection-in-industry-manufacturing/tree/master/anomalib_contribute

Technology used

App Flask

python app.py

Open local URL: http://127.0.0.1:5000

Default account login:

• Username: anomalib
• Password: anomalib

Models supported:

Login:

#:

Homepage:

Train:

Inference:

Database:

Container

docker build -t anomaly:v1 .
docker run anomaly:v1

or just simple

docker-compose up

or use my docker image in Docker hub (Coming soon)

Deploy AWS

First: Create EC2 instance

Second: git clone and install related packages

git clone https://github.com/vnk8071/anomaly-detection-in-industry-manufacturing.git

Next: install Miniconda and Docker engine

• Miniconda: Follow link https://varhowto.com/install-miniconda-ubuntu-18-04/
• Docker engine: Follow link https://docs.docker.com/engine/install/ubuntu/

docker-compose up

Final: access link http://user-IPv4-public-ec2-aws