SimeonHristov99/ViZTranZ

A Visual Translation Service that leverages AWS services.

ViZTranZ

Goal: Upload an image and get translation of the objects in the image.

Quick Start

./setup.sh
./run.sh

Tasks

• Create local project.
• Basic UI.
• Create github project.
• Create AWS account.
• Set up S3 bucket that would act as a trigger to the lambda.
• Create the AWS Lambda function.
• Set up AWS Rekognition.
• Connect lambda to another S3 to publish results.
• Prettify code.
• Set up AWS Translate.
• Add sidemenu with AWS and Local modes.
• Link Hugging Face to UI.
• Finish Local mode.
• Package code.
• Add bash script setup.sh instead of commands in quick start section.
• Add bash script run.sh.
• Add versions to the pip packages.
• Add logo!
• Upload paper (in bulgarian).

About

Created as part of a university course: Application programming interfaces for Cloud Architectures with AWS.

References

• AWS Tutorial : Image recognition and notification with AWS Lambda, Rekognition, SNS, S3, Python: https://www.youtube.com/watch?v=wnTvVB1ojPk
• AWS: Image analysis using Rekognition via Lambda function: https://www.youtube.com/watch?v=3r_ue7TQkCE
• Using AWS Rekognition and Python to Identify Objects / Text in Images: https://www.youtube.com/watch?v=gjUwPEqnEeI
• AWS Architecture: Serverless Photo Recognition: https://www.youtube.com/watch?v=GIdJz7VnP58
• Translating documents with Amazon Translate, AWS Lambda: https://www.youtube.com/watch?v=-_2wCN5heXw&t=448s
• Tensorflow Hub model for object detection: https://tfhub.dev/google/openimages_v4/ssd/mobilenet_v2/1
• Hugging Face model for translating en-de: https://huggingface.co/Helsinki-NLP/opus-mt-en-de
• Hugging Face model for translating en-bg: https://huggingface.co/Helsinki-NLP/opus-mt-en-bg
• Hugging Face model for translating en-ru: https://huggingface.co/Helsinki-NLP/opus-mt-en-ru