In-Car-Abandoned children’s body temperature detection based on thermal imaging using deep learning approaches / Muhammad Mohamad Gzazali

The term “In-Car-Abandoned Children” refers to children left unsupervised in vehicles. The number of child deaths due to being trapped inside vehicles is increasing every day. Recent technology has introduced many sensors for detecting in-car-abandoned children, primarily focusing on image and sound detection, which can be inaccurate. This study proposes a sensor based on thermal body of children an adding detection element to improve accuracy. The child's body temperatures will be classified using deep learning approaches, specifically employing Python for data acquisition and system refinement. The tuned Python model aims to achieve optimal results. The findings demonstrate the efficacy of the proposed deep learning model, specifically the MobileNet V2 model, achieving remarkable classification accuracies: 97% for compact cars, 95% for MPVs, and 93% for sedans. These results underscore the model's capability in accurately identifying children abandoned in vehicles using thermal images. The optimized models significantly enhanced accuracy, reinforcing the model's reliability across various car types. This study is significant due to its demonstrated accuracy and stability in detecting children in vehicles. By proposing an alternative method for identifying abandoned children in cars using both low-quality and high-quality thermal images, this research provides a valuable tool for authorities. It offers a practical solution to mitigate the risks associated with child heatstroke in vehicles. Ultimately, the successful implementation of this model has the potential to substantially reduce the incidence of child abandonment in cars and promote safer transportation practices for children. This research supports authorities, particularly the Ministry of Transportation, in risk mitigation. Successful implementation could reduce child abandonment and promote safer child transportation.