Smart Home Danger Detection + Alert

Final-project-in-progress for 17-422/722,05-499/899, Spring 2024: Building User-Focused Sensing Systems.
This code is built upon official repositories of CLAP and ViFi-CLIP.

Environment Setup

This environment has been tested on MacBook Pro with M1 Pro chip.

conda env create -f environment.yaml

Download Checkpoints

• For CLAP, weights will be downloaded automatically.
• For Video Finetuned CLIP, go to ./ViFi-CLIP and create a new folder named ckpts. Download this checkpoint and move it to ./ViFi-CLIP/ckpts.

Run Inferences

• To run CLAP inference, open clap_inference.ipynb. Specify label csv file path and video path in the first cell and run all.
• To run Video Finetuned CLIP inference, open ViFi-CLIP_inference.ipynb. Specify label csv file path and video path in the first cell and run all.

Run Pi Code

In the smarthome-pi folder, run python3 main.py

Run Webapp

In the webapp folder, run

python3 manage.py makemigrations
python3 manage.py migrate
python3 manage.py runserver