MMLAPipeline-Resources

Demos, posters and other resources related to the ChimeraPy MMLA Pipeline ecosystem for EngageAI Institute site visit 2023.

Posters

1. ChimeraPy and the MMLAPipeline

Demos

Multi-camera Streaming

multi-camera.json includes the ChimeraPyOrchestrator config streaming multiple web cameras. This requires 2 external webcams (along with the native webcams) connected to a linux and windows machine. In the manager, we also run a webrtc-chimerapy-node.

The architecture is as follows:

multi-camera.mp4

Gaze Detection with Screen Capture

gaze-camera-screen.json is the ChimeraPyOrchestrator configuration for capturing screen and webcam images on a remote machine, feeding it to a gaze detection network (L2CS-Net) and displaying results in a local machine. The architecture is as follows:

gaze-demo.mp4

Multi-video Tracking

multi-video-tracking.json is the ChimeraPyOrchestrator configuration for tracking multiple objects in multiple videos. The architecture is as follows:

multi-video-tracking.mp4

Embodied Learning

embodied-learning.json is the ChimeraPyOrchestrator configuration for embodied learning. The architecture is as follows:

embodied-learning.mp4

Useful Links

• ChimeraPy: Distributed computing framework for Multimodal data written in Python

• ChimeraPyOrchestrator: Reusable Nodes and Orchestration Scheme/ Dashboard Application for ChimeraPy with JSON configuration

• MMLAPIPE: Repository of shareable ChimeraPy pipelines