Model Card for Snow-Cover and Visibility Estimation using ResNet50

This ResNet50 model is trained to provide a relative estimation of snow cover and horizontal visibility from webcam images. The model outputs are averaged to provide proxies for these two weather-related metrics.

For details and a quick starter Colab, visit GitHub repository.

Model Details

Model Description

This model leverages a ResNet50 architecture to estimate snow cover and visibility based on webcam imagery. It was trained using a dataset of images captured in varying weather and lighting conditions, making it suitable for outdoor use.

Developed by: Pierre Lepetit
Funded by: Météo-France
Model type: Convolutional Neural Network (ResNet50)
Language(s): N/A (not applicable for this non-NLP model)
License: CC by 4.0
Training data: Webcam images from various outdoor environments, annotated for snow cover and visibility.

Input and Output

Input: Webcam images (standard format, JPEG/PNG, etc.) of outdoor environments.
Output:
- Relative snow cover estimation.
- Relative Horizontal visibility estimation.

Uses

Intended Use

This model is designed for:

Monitoring snow cover in outdoor environments.
Estimating visibility conditions in real-time based on webcam imagery.

Direct Use

Refer to the example notebook for direct use: example_usage.ipynb