An Effective AI System for Accurate Segmentation of the Hybrid Visual Character Analysis of a Coral Bleaching Domain

Coral reefs, which are marine gardens that contain immobile invertebrates, are at risk from a concerning phenomenon known as coral bleaching, which is caused by global warming. Such pale corals, which provide refuge for a vast array of marine species, particularly impact marine pharmacognosy. With regards to this matter, numerous researchers are diligently seeking remedies and therapeutic methods to recover the bleached corals. Nevertheless, a substantial effort is required to address this issue, which including accurate detection and cartography of bleached coral regions inside extensive reef networks like the Great Barrier Reef (GBR). Numerous academics have put up several visual classification paradigms to precisely classify corals as bleached. Nevertheless, these approaches are somewhat less efficient when faced with variables such as fluctuations in light intensity, perspectives, dimensions, and locations. This work introduces a very resistant and uncompromising localisation technique known as RL-BoHVF (Resolute Localisation utilising Bag-of- Hybrid Visual Features) to address these tasks. RL-BoHVF blends the capabilities of deep neural networks by leveraging the deeper neural architecture of AlexNet, while incorporating manually designed features derived from the main input, such as ColorTexture. These achievements are attained by implementing the bag-of-feature technique, which distinguishes RL-BoHVF as effective in minimising the intricate feature structure while simultaneously achieving a remarkable categorisation accuracy of 96%. 20% on a meticulously calibrated sample derived from the Great Barrier Reef. Furthermore, the suggested model's objectivity was meticulously calibrated by ensuring both computational efficiency and accuracy in identification and recognition. This was achieved by using 342 photos in the combined training and validation segments. Empirical evidence has confirmed that the suggested model, RL-BoHVF, surpasses all other autonomous and composite models, including those developed using deep neural networks (DNN) and those created manually and documented in previous studies. These findings provide evidence for the utility and efficacy of the described method in identifying the precise locations of bleached corals, thereby enhancing conservation initiatives in coral reef environments.