Artificial Intelligence Based Object Detection and Tracking for a Small Underwater Robot

Object recognition and tracking is a challenge for underwater vehicles. Traditional algorithm requires a clear feature definition, which suffers from uncertainty as the variation of occlusion, illumination, season and viewpoints. A deep learning approach requires a large amount of training data, which suffers from the computation. The proposed method is to avoid the above drawbacks. The Siamese Region Proposal Network tracking algorithm using two weights sharing is applied to track the target in motion. The key point to overcome is the one-shot detection task when the object is unidentified. Various complex and uncertain environment scenarios are applied to evaluate the proposed system via the deep learning model's predictions metrics (accuracy, precision, recall, P-R curve, F1 score). The tracking rate based on Siamese Region Proposal Network Algorithm is up to 180 FPS.

Automated summary

Object recognition and tracking in underwater vehicles pose challenges. Traditional algorithms with clear feature definitions suffer from uncertainties caused by occlusions, illumination changes, seasonal variations, and different viewpoints. Deep learning approaches require extensive training data but face computation challenges. To overcome these drawbacks, the proposed method utilizes the Siamese Region Proposal Network tracking algorithm with weight sharing to track moving targets. The key challenge is the one-shot detection task for unidentified objects. The proposed system is evaluated using complex and uncertain environmental scenarios, with deep learning model metrics such as accuracy, precision, recall, P-R curve, and F1 score. The tracking rate based on the Siamese Region Proposal Network Algorithm can reach up to 180 FPS.