Human Activity Recognition (HAR) using 1D CNN | Hugging Face Deployment | EG20225032 скачать в хорошем качестве

Human Activity Recognition (HAR) using 1D CNN | Hugging Face Deployment | EG20225032

Machine Learning Project – Human Activity Recognition (HAR) using Wearable Sensors In this video, I demonstrate my sequence-based HAR project using accelerometer data from wearable sensors. The model classifies six activities: Cycling, Push-up, Run, Squat, Table Tennis, and Walk. Workflow covered in the demo: Data loading and cleaning Sequence windowing with K = 30 consecutive samples Train/validation/test split (70% / 10% / 20%) Feature normalization using a saved scaler Training a 1D CNN time-series classifier Evaluation using accuracy, classification report, and confusion matrix Deployment on Hugging Face Spaces (Gradio) with CSV upload and prediction + probabilities Hugging Face Space: https://huggingface.co/spaces/emras21/EG20... Student ID: EG20225032 Name: Prameeth Evans Course: EE5203-24