metadata

title: Submission Template
emoji: 🔥
colorFrom: yellow
colorTo: green
sdk: docker
pinned: false

MountAIn model for smoke detection

Model Description

This is an evolution from YOLO baseline to focus on small-to-medium objects and integrated SAHI-like approach

Intended Use

Primary intended uses: First submission of a novel class model
Primary intended users: Researchers and developers participating in the Frugal AI Challenge
Out-of-scope use cases: Not intended for production use or real-world classification tasks

Training Data

The model the Pyro-SDIS Subset contains 33,636 images, including:

28,103 images with smoke
31,975 smoke instances

Labels

Smoke

Performance

Metrics

Accuracy: Still to be estimated but mAP:50 > 70%
Environmental Impact:

Emissions impact if inference is run on Cloud and/or on-premise gateways
- Emissions tracked in gCO2eq
- Energy consumption tracked in Wh
Emissions are null if run on MountAIn vision sensors since they are powered by renewable energy

Model Architecture

Evolution from YOLO baseline

Environmental Impact

Environmental impact is tracked using CodeCarbon, measuring:

Carbon emissions during inference
Energy consumption during inference

This tracking helps establish a baseline for the environmental impact of model deployment and inference while running in Cloud and/or on-premise gateways. The usage of MountAIn vision sensors enables no environmental impact thanks to the usage of renewable energy

Limitations

Not suitable for any real-world applications as is without proper export to tiny MCUs

Ethical Considerations

Environmental impact is tracked to promote awareness of AI's carbon footprint