Overview

Halo Solutions aimed to enhance their video recording and processing capabilities by integrating advanced video streaming and AI-powered analytics. The objective was to enable real-time video processing and analysis using AWS and Google Cloud technologies, providing a seamless and scalable solution for their clients.

Architecture and Integration

The solution architecture involved a combination of AWS services and Google Cloud Vision, facilitating a comprehensive video processing application:

• AWS IVS (Interactive Video Service): Enabled real-time video streaming for interactive experiences within applications.
• Kinesis Video Streams (KVS): Used to ingest and store live video streams from a single camera (referred to as a zone), enabling secure video data handling.
• AWS Rekognition: Deployed for real-time video analysis, including dynamic head count, density, and overall sentiment of the crowd.
• Google Cloud Vision: Enhanced AWS Rekognition by providing additional AI-driven image and video analysis.
• AWS CloudFormation: Automated the deployment of AWS resources, reducing manual intervention and errors.
• ReactJS: Integrated as the front-end framework for building interactive dashboards for video management and analytics.

Key Features and Solutions

Key features of the integration include:

• Real-Time Video Streaming: AWS IVS provided low-latency, high-quality live video streaming capabilities, suitable for live broadcasts and interactive applications.
• Dynamic Video Analysis: AWS Rekognition was used for detecting faces, counting heads, and analyzing crowd sentiment dynamically.
• Scalable Video Storage: Kinesis Video Streams facilitated the storage and management of large volumes of video data with high availability and durability.
• Automated Resource Deployment: AWS CloudFormation was employed to automate resource provisioning, ensuring consistent deployment and reducing setup time.
• User Interface for Video Management: ReactJS was used to create a user-friendly interface for managing video streams and analyzing data.

Challenges and Solutions

Several challenges were addressed during the integration process:

• Latency in Video Processing: Potential latency issues in video streaming and processing were mitigated by optimizing AWS IVS configurations and using AWS Lambda for edge processing.
• Data Security and Privacy: Ensured compliance with data protection regulations through AWS IAM roles, VPCs, and encryption for video data.
• Integration with AI Tools: Integrated AWS Rekognition and Google Cloud Vision using a unified API layer to handle multiple AI outputs and provide consistent analytics.

Proposed Solutions & Issues

To overcome the streaming video issue, the following solution was implemented:

• Storing Images from Video Streams: Kinesis Video Streams was configured to periodically store images in an S3 bucket for further analysis, leveraging AWS Rekognition for detecting emotions and other parameters.
• Challenges with GStreamer Plugin: The GStreamer plugin required modifications in the CPP source code to add tags for storing images, which presented some difficulties in integration.

Recommendations and Future Enhancements

• Expand AI capabilities by integrating additional machine learning models for advanced video analysis and predictions.
• Enhance the user interface to include more customizable analytics dashboards and reporting features.
• Implement AWS Elemental MediaConvert for advanced video editing and transcoding, providing more flexibility in video content delivery.