This project runs a activity classification model (Random Forest/Decision Tree) on the cloud based on sensors deployed in a house (ARUBA DATASET) and it predicts based on that. The actual setup of the sensors are not done, so it’s a simulated sensor setup that randomly takes sensor values and according to timestamps and sends to the model to classify and hence predict the activity done at that time using those virtual sensor data. This project serves as a proof of concept that one can predict and monitor activity over the internet through sensors while being away for tasks such as elderly care monitoring

The proposed model in this paper uses an improved architecture of YOLOv7-Tiny. The deployment of the defect detection model on a Raspberry Pi allows for real-time detection and remote monitoring of welding defects. This technology can be particularly useful in hazardous and remote locations where manual inspection is difficult or impossible. The implementation of this model can lead to significant cost savings, improved safety, and increased productivity for industrial operations.