Human Action Recognition using Fuzzy Logic and VGG16 Classifier

Human action recognition (HAR) is a challenging task with applications in various domains such as surveillance, healthcare, and human-computer interaction. This paper aims to develop a HAR system using a combination of fuzzy logic and the VGG16 convolutional neural network (CNN). The proposed approach begins with the collection and preparation of a labeled dataset consisting of images representing different human activities. Preprocessing techniques are applied to enhance the quality and consistency of the images. Fuzzy logic is then utilized to predict the membership values for each action and perform fuzzy membership calculations on the training labels, leveraging its ability to capture ambiguity and uncertainty associated with human activities. The VGG16 CNN architecture, known for its effectiveness in image classification, is employed for feature extraction and classification. A pre-trained VGG16 model is finetuned on the HAR dataset, replacing the last layers to adapt it to the specific task. The model is trained using a training set while monitoring its performance on a validation set to optimize hyperparameters and ensure the model’s robustness. By combining fuzzy logic and the VGG16, this proposed HAR system, integrating fuzzy logic and the VGG16 CNN, exhibits promising potential for accurately recognizing human activities in images.