STP shoplifting Competition

I participated in the STP Shoplifting Detection Competition, where our team ranked among the top five teams. Along with my teammates, Amjad Bakri and Abdelrhman Wael, we designed and implemented an AI-based shoplifting detection system using deep learning techniques. The project focused on analyzing video data to identify suspicious behavior in retail environments. We experimented with both pre-trained and non-pre-trained models and evaluated multiple architectures, including ConvLSTM, LRCN, Inception, MobileNet, and other deep learning models. Through extensive experimentation, we concluded that transformer-based models are among the most effective approaches for this problem; however, due to competition constraints such as limited processing time, we were unable to deploy them in the final submission. This experience strengthened my practical skills in computer vision, model selection, performance optimization, and working under real-world constraints in AI competitions.