Title

AWS re:Invent 2022 - Leveling up computer vision and artificial intelligence development (ANT218)

Summary

• Dave Malone, an IoT Specialist Solutions Architect at AWS, and Dr. Ajay Gulati, CTO of Kami Vision, presented on leveraging Amazon Kinesis Video Streams for computer vision applications.
• They discussed the use of connected video in various sectors, including smart homes, industrial monitoring, public safety, and healthcare.
• Challenges in building connected camera solutions include scalability and security.
• Kinesis Video Streams simplifies streaming video data from devices to AWS for analytics, machine learning, and other processing.
• Two streaming methods are available: Kinesis Video Streams for unidirectional streaming and Kinesis Video Streams WebRTC for bidirectional, low-latency streaming.
• Producer SDKs in C++, C, and Java are available on GitHub to help developers get started with streaming.
• Kinesis Video Streams offers features like media storage, playback, indexing, encryption, and integration with other AWS services like SageMaker and Amazon Rekognition.
• Dr. Gulati presented Kami Vision's use case, KamiCare, a fall detection solution for elder care using the AWS platform.
• KamiCare leverages AI models on cameras for accurate fall detection and quick alerting to care staff.
• The platform supports various cameras, device management, event handling, storage, user access control, and notifications.
• Benefits of using AWS services include reduced development time, no need for dedicated video management teams, and scalability without DevOps concerns.
• AWS provided access to certified developers and partners, financial incentives, and support.
• Kami Vision also built a Personal Emergency Response System (PERS) with Bay Alarm Medical using the platform.
• The session concluded with a live demo of KamiCare and a discussion on the benefits of Kinesis Video Streams.

Insights

• The use of AWS services like Kinesis Video Streams can significantly accelerate the development of complex video-based solutions by providing ready-to-use infrastructure and SDKs.
• The ability to run AI models on camera devices can reduce latency and bandwidth requirements, making real-time applications like fall detection more efficient and cost-effective.
• AWS's ecosystem, including its certified developers and partner network, can be leveraged to enhance development capabilities and reduce time-to-market for new products.
• Real-world deployment of solutions often reveals unforeseen challenges, emphasizing the importance of field testing and the ability to quickly iterate on product features.
• The flexibility of AWS services allows for the creation of both specific solutions like KamiCare and more generic platforms that can be adapted or white-labeled for various use cases.
• AWS's support structure and incentives for startups and partners can be a significant advantage for companies looking to innovate with AWS technologies.