ruview-model-training

What problem does it solve?

This Skill addresses the need for training and deploying various RuView models with advanced features like domain generalization and on-device SNN environment adaptation.

Core Features & Use Cases

• Model Training: Offers multiple tracks for camera-free, camera-supervised, and RuVector contrastive embeddings models.
• Domain Generalization: Allows model transfer across environments without retraining.
• SNN Adaptation: Provides local SNN environment adaptation for efficient training.
• GPU Training: Facilitates GPU training on GCloud and Hugging Face publishing for models.
• Use Case: Ideal for developers looking to build, fine-tune, evaluate, or ship advanced RuView models for applications requiring robust pose estimation and domain adaptation.

Quick Start

Train a camera-free pose model using the 'wifi-densepose-sensing-server' with the command: cargo run -p wifi-densepose-sensing-server -- --pretrain --dataset data/csi/ --pretrain-epochs 50.