Human Activity Recognition With Open CV and Deep Learning

  • Unique Paper ID: 174815
  • Volume: 11
  • Issue: 11
  • PageNo: 638-642
  • Abstract:
  • Human Activity Recognition (HAR) plays a crucial role in various applications such as surveillance, healthcare, and human-computer interaction. In this paper, we propose a new HAR system via Convolution Neural Networks (CNN) which is one of deep learning algorithms. The proposed system encompasses a comprehensive dataset comprising 400 diverse human activities, enabling a robust and versatile model. Leveraging the power of OpenCV (Open Source Computer Vision Library) and Deep Learning techniques, the system aims to automatically identify and classify various human actions in real-time. The dataset encompasses a wide spectrum of activities, ensuring the model's ability to generalize across diverse scenarios. The 3D CNN model used in this model is RESNET-34. The activities include both routine daily tasks and anomalous behaviors, providing the system with the capacity to identify unusual or suspicious actions. The extensive dataset also contributes to the adaptability of the model in different environments and cultural contexts. The process involves capturing video data, preprocessing it using OpenCV to extract relevant features, and feeding these features into a Deep Learning model. The model, trained on a dataset of diverse human activities, learns to associate specific patterns with different actions. This enables the system to predict and classify ongoing activities, ranging from simple gestures to complex movements. We have achieved the average recognition accuracy of 89.99% for the activities.

Cite This Article

  • ISSN: 2349-6002
  • Volume: 11
  • Issue: 11
  • PageNo: 638-642

Human Activity Recognition With Open CV and Deep Learning

Related Articles