arXiv reaDer
Underwater litter monitoring using consumer-grade aerial-aquatic speedy scanner (AASS) and deep learning based super-resolution reconstruction and detection network
Underwater litter is widely spread across aquatic environments such as lakes, rivers, and oceans, significantly impacting natural ecosystems. Current monitoring technologies for detecting underwater litter face limitations in survey efficiency, cost, and environmental conditions, highlighting the need for efficient, consumer-grade technologies for automatic detection. This research introduces the Aerial-Aquatic Speedy Scanner (AASS) combined with Super-Resolution Reconstruction (SRR) and an improved YOLOv8 detection network. AASS enhances data acquisition efficiency over traditional methods, capturing high-quality images that accurately identify underwater waste. SRR improves image-resolution by mitigating motion blur and insufficient resolution, thereby enhancing detection tasks. Specifically, the RCAN model achieved the highest mean average precision (mAP) of 78.6% for detection accuracy on reconstructed images among the tested SRR models. With a magnification factor of 4, the SRR test set shows an improved mAP compared to the conventional bicubic set. These results demonstrate the effectiveness of the proposed method in detecting underwater litter.
updated: Wed Aug 07 2024 05:56:05 GMT+0000 (UTC)
published: Wed Aug 07 2024 05:56:05 GMT+0000 (UTC)
参考文献 (このサイトで利用可能なもの) / References (only if available on this site)
被参照文献 (このサイトで利用可能なものを新しい順に) / Citations (only if available on this site, in order of most recent)
Amazon.co.jpアソシエイト