Preliminary Study on the Detection of Subtle Variations in Image Sequences for Identifying False Starts in Speedway Racing

Computer Vision algorithms have gained significant popularity and prove to be highly beneficial in detecting motion. These methods have proven to be profitable in motor sports especially. This paper presents a preliminary study on image processing, computer vision, and Long Short-Term Memory (LSTM) networks to detect subtle variations in image sequences for identifying false starts in speedway racing. Traditional methods often rely on manual observation, which lacks the precision to capture delicate movements at the start line. The presented methodology utilizes mentioned methods to enhance image quality, computer vision to detect and track racers’ positions, and LSTM networks to analyze temporal sequences for movements indicating a false start. Results indicate that the adopted approach outperforms manual false start detection in accuracy and reliability. This advancement not only ensures fairness in speedway racing by providing a more objective detection of false starts but also demonstrates the potential of these technologies for applications requiring the identification of subtle changes in sequences. The efficiency and precision of our method offer promising implications for sports regulations and beyond, advocating for the broader application of technology in similar contexts.