To Build and Evaluate Irregular Human Activity Detection System Using Deep CNN-LSTM Methodology

The abstract of this model is a study into irregular human activity identification, which is an important requirement for automated surveillance systems that can best match and reliably identify irregular activities. The work presented a novel and efficient method for evaluating sequential data gathered from photos by merging CNNs with LSTMs. Methods were created to properly preprocess image and video-based data, train convolutions to extract discriminative features, integrate LSTMs for temporal dependency capture, and evaluate the methodology's performance on a variety of video and picture datasets. In this research taken dataset from the UCF crime dataset based on the real world anomaly detection in surveillance videos where The dataset contains images extracted from every video from the UCF Crime Dataset. Every 10th frame is extracted from each full-length video and combined for every video in that class. All the images are of size 64*64 and in. png format. The dataset has a total of 14 Classes. The CNN-LSTM technique increased abnormal occurrence identification with significantly higher accuracy and strength across a wide range of abnormal behavior patterns, including rapid and irregular body movements, object contact, and travel behavior. The unique integration of physical and temporal data representation in this work has significant ramifications for bolstering security systems, health monitoring, and other real-world applications requiring automated irregular detection.