AI-Based Cameras to Detect a Leak of Chemicals or Gas in Real Time
P. Selvakumar, M. Sivaraja, D. Satishkumar, Ganesh Vijayakumar, K. Satheesh Kumar, B. S. Navaneeth
Abstract
In many real-world applications, strong and dependable detection depends on sensor fusion, as single-sensor sensing may not be sufficient. The individual modality feature extraction module is part of the network design. It is fused using a merging layer and a dense layer to produce a single output for gas identification. Using the fused model, the authors were able to achieve a testing accuracy of 96%, compared to the 82% and 93% accuracy of the individual models, respectively, based on gas sensor data using LSTM and thermal representation data using CNN models. The combined use of numerous sensors and modalities yields better results than using just one sensor. Since most of the gases and their vapors are tasteless, odorless, and colorless, they interfere with our ability to perceive things normally.