The detection and preservation of marine biodiversity has garnered global attention. The incorporation of deep learning methodologies can elevate the efficiency of species detection. In this study, we developed a DeepSeaNet for effective localization and accurate classification of organisms based on deep-sea images, as well as for hinting at unknown organisms (new species). The DeepSeaNet fully accommodates the unique characteristics of deep-sea organisms and imaging environment, leading to remarkable advancements in fine-grained analysis and accuracy. The DeepSeaNet comprises two network components: a deep-sea Classes Detection Network (CDN) and an unsupervised Species Clustering Network (SCN). CDN is used for biological class detection and is specifically tailored for deep-sea environments. It incorporates modules for feature fusion, multi-scale analysis, and self-attention. SCN is specifically designed to detect and identify new species by utilizing the location information extracted from the CDN output results. It is composed of a feature extraction module and a clustering module. By collecting deep-sea image data from the “KeXue” Science Research Vessel, we constructed a dataset totaling 29,436 images of deep-sea organisms covering more than 500 species of deep-sea seamount organisms. This dataset serves as the foundational dataset for our experiment. As a result, our model achieves an 82.18% mean average precision for class detection and a 43.4% accuracy for species detection. Furthermore, the model has the capability to identify new species through the computation of inter-species distances.

Object detection algorithms, a popular research direction in computer vision, are valued for their wide range of applications. However, its applications in marine deep-sea biology are rarely seen. Three reasons are hidden among them: first, the lack of marine deep-sea datasets; second, the difficulty of network detection units to meet taxonomic requirements (both in speed and accuracy); and third, the difficulty of judging the network as a whole. To solve these problems to achieve high accuracy and high fineness for deep-sea target detection tasks, this paper proposes a creature detection model (CDM) based on YOLO V7 tiny network incorporating the idea of a bag of features. The network proposes a complementary LOSS function to optimize the model error caused by the polarization of underwater robots (ROVs) during deep-sea operations and uses an unsupervised clustering algorithm to generate a new dictionary for underwater species classification based on the detection.

SUMMARY: iTOL is a powerful and comprehensive phylogenetic tree visualization engine. However, adjusting to new templates can be time-consuming, especially when many templates are available. We developed an R package namely itol.toolkit to help users generate all 23 types of annotation files in iTOL. This R package also provides an all-in-one data structure to store data and themes, accelerating the step from metadata to annotation files of iTOL visualizations through automatic workflows. AVAILABILITY AND IMPLEMENTATION: The manual and source code are available at https://github.com/TongZhou2017/itol.toolkit.