How Veritone uses Amazon Bedrock, Amazon Rekognition, Amazon Transcribe, and information retrieval to update their video search pipeline

This article discusses how Veritone, an AI company, uses various AWS services to improve their video search pipeline by incorporating semantic retrieval based on text queries. The key points are:

Specifically, the article covers:

• Solution overview using AWS services like Amazon Bedrock, Amazon Rekognition, Amazon Transcribe, Amazon Comprehend, and Amazon OpenSearch Service
• Details of the metadata generation pipeline for processing videos and generating embeddings
• The search pipeline with indexing, query processing, and result combination strategies
• Evaluation pipeline with a UI for qualitative evaluation of search methods
• Experiments and results on short and long video datasets, showcasing the improvements achieved with semantic retrieval methods
• Key takeaways and security best practices