Improving accuracy and efficiency in seagrass detection using state-of-the-art AI techniques

Seagrasses provide a wide range of ecosystem services in coastal marine environments. Despite their ecological and economic importance, these species are declining because of human impact. This decline has driven the need for monitoring and mapping to estimate the overall health and dynamics of seagrasses in coastal environments, often based on underwater images. However, seagrass detection from underwater digital images is not a trivial task; it requires taxonomic expertise and is time-consuming and expensive. Recently automatic approaches based on deep learning have revolutionised object detection performance in many computer vision applications, and there has been interest in applying this to automated seagrass detection from imagery. Deep learning–based techniques reduce the need for hardcore feature extraction by domain experts which is required in machine learning-based techniques. This study presents a YOLOv5-based one-stage detector and an EfficientDetD7–based two-stage detector for detecting seagrass, in this case, Halophila ovalis, one of the most widely distributed seagrass species. The EfficientDet-D7–based seagrass detector achieves the highest mAP of 0.484 on the ECUHO-2 dataset and mAP of 0.354 on the ECUHO-1 dataset, which are about 7% and 5% better than the state-of-the-art Halophila ovalis detection performance on those datasets, respectively. The proposed YOLOv5-based detector achieves an average inference time of 0.077 s and 0.043 s respectively which are much lower than the state-of-the-art approach on the same datasets.     

A case study of utilizing YOLOT based quantitative detection algorithm for marine benthos

Current object detection algorithms suffer from low accuracy and poor robustness when used to detect marine benthos due to the complex environment and low light levels on the seabed. To solve these problems, the YOLOT (You Only Look Once with Transformer) algorithm, a quantitative detection algorithm based on the improved YOLOv4, is proposed for marine benthos in this paper. To improve the feature extraction capability of the neural network, the transformer mechanism is introduced in the backbone feature extraction network and feature fusion part of YOLOv4, which enhances the adaptability of the algorithm to targets in complex undersea environments. On the one hand, the self-attention unit is embedded into CSPDarknet-53, which improves the feature extraction capability of the network. On the other hand, it is transformer-based feature fusion rules that are introduced to enhance the extraction of contextual semantic information in the feature pyramid network. In addition, probabilistic anchor assignment based on Gaussian distribution is introduced to network training. The experimental validation shows that compared with the original YOLOv4, the YOLOT algorithm improves the recognition precision from 75.35% to 84.44% on the marine benthic dataset. The improvement reflects that YOLOT is suitable for the quantitative detection of marine benthos.     

WilDect-YOLO: An efficient and robust computer vision-based accurate object localization model for automated endangered wildlife detection

Objective. With climatic instability, various ecological disturbances, and human actions threaten the existence of various endangered wildlife species. Therefore, an up-to-date accurate and detailed detection process plays an important role in protecting biodiversity losses, conservation, and ecosystem management. Current state-of-the-art wildlife detection models, however, often lack superior feature extraction capability in complex environments, limiting the development of accurate and reliable detection models. Method. To this end, we present WilDect-YOLO, a deep learning (DL)-based automated high-performance detection model for real-time endangered wildlife detection. In the model, we introduce a residual block in the CSPDarknet53 backbone for strong and discriminating deep spatial features extraction and integrate DenseNet blocks to improve in preserving critical feature information. To enhance receptive field representation, preserve fine-grain localized information, and improve feature fusion, a Spatial Pyramid Pooling (SPP) and modified Path Aggregation Network (PANet) have been implemented that results in superior detection under various challenging environments. Results. Evaluating the model performance in a custom endangered wildlife dataset considering high variability and complex backgrounds, WilDect-YOLO obtains a mean average precision (mAP) value of 96.89%, F1-score of 97.87%, and precision value of 97.18% at a detection rate of 59.20 FPS outperforming current state-of-the-art models. Significance. The present research provides an effective and efficient detection framework addressing the shortcoming of existing DL-based wildlife detection models by providing highly accurate species-level localized bounding box prediction. Current work constitutes a step toward a non-invasive, fully automated animal observation system in real-time in-field applications.     

Cascaded deep network systems with linked ensemble components for underwater fish detection in the wild

We propose a fish detection system based on deep network architectures to robustly detect and count fish objects under a variety of benthic background and illumination conditions. The algorithm consists of an ensemble of Region-based Convolutional Neural Networks that are linked in a cascade structure by Long Short-Term Memory networks. The proposed network is efficiently trained as all components are jointly trained by backpropagation. We train and test our system for a dataset of 18 videos taken in the wild. In our dataset, there are around 20 to 100 fish objects per frame with many fish objects having small pixel areas (less than 900 square pixels). From a series of experiments and ablation tests, the proposed system preserves detection accuracy despite multi-scale distortions, cropping and varying background environments. We present analysis that shows how object localization accuracy is increased by an automatic correction mechanism in the deep network's cascaded ensemble structure. The correction mechanism rectifies any errors in the predictions as information progresses through the network cascade. Our findings in this experiment regarding ensemble system architectures can be generalized to other object detection applications.     

An image enhancement approach for coral reef fish detection in underwater videos

Coral reefs are rich in fisheries and aquatic resources, and the study and monitoring of coral reef ecosystems are of great economic value and practical significance. Due to complex backgrounds and low-quality videos, it is challenging to identify coral reef fish. This study proposed an image enhancement approach for fish detection in complex underwater environments. The method first uses a Siamese network to obtain a saliency map and then multiplies this saliency map by the input image to construct an image enhancement module. Applying this module to the existing mainstream one-stage and two-stage target detection frameworks can significantly improve their detection accuracy. Good detection performance was achieved in a variety of scenarios, such as those with luminosity variations, aquatic plant movements, blurred images, large targets and multiple targets, demonstrating the robustness of the algorithm. The best performance was achieved on the LCF-15 dataset when combining the proposed method with the cascade region-based convolutional neural network (Cascade-RCNN). The average precision at an intersection-over-union (IoU) threshold of 0.5 (AP₅₀) was 0.843, and the F1 score was 0.817, exceeding the best reported results on this dataset. This study provides an automated video analysis tool for marine-related researchers and technical support for downstream applications.     

Introduction to fish imagery in art

Synopsis Fish have been the subject of works of art for at least 14000 years and appeared in primitive art from many cultures. In ancient civilizations of the West, fishes were a constant, if infrequent, motif. Fish designs in ancient Egypt were common and showed little change for 1500 years. Decorative fish designs of the Greeks and Romans (often with mythological significance) were adopted by early Christians as religious symbols. With the development of printing, the non-religious depiction of fish became more widespread and realistic paintings of fish, especially still lifes, appeared during the Renaissance. This still life tradition reached a peak in 17th century Netherlands. After 1750, fish images appeared in many different contexts. Realistic painters showed the agony of newly-caught fish, dramatic marine scenes with fish, and occasionally freshwater fishes in their habitats. In the twentieth century, fish were painted by many modern artists, including Matisse, Picasso, Klee, Masson, Beckman, Soutine, Magritte, and Thiebaud. Some of these artists' fish images are pleasing, others are violent or ambiguously symbolic. In contrast, contemporary nature artists tend to paint live fish in idealized settings, a style with roots in 17th century still lifes and oriental brush paintings. In Japan and China, fish have been an important theme in art and their use has been highly symbolic. A survey of fish images in art shows that artists, like scientists, create mainly in the context of historical precedents.     

Natural variations in estuarine fish,fish odor,and zooplankton photobehavior

ABSTRACT

Fish odor induces predator avoidance behaviors in zooplankton, like vertical migration, by making zooplankton more responsive to light. Odor cues that alter behavior in marine crustacean zooplankton in the laboratory include sulfated glycosaminoglycans (sGAGs) derived from fish body mucus. Few studies quantify these cues in estuarine/marine environments or assess whether laboratory studies reflect natural scenarios. We collected fish and water samples weekly in Broadkill River, Delaware, USA. We used field-collected water in colorimetric assays to determine the concentration of sGAG-equivalent molecules and in behavioral assays with a zooplankton model, brine shrimp (Artemia spp.) nauplii, which only descend in response to downwelling light after fish odor exposure. Fish quantity was positively related to sGAG-equivalents and zooplankton photosensitivity, indicated by descent responses at lower light levels and across a broad intensity range. Our results support that fish odor concentrations used in previous laboratory assays are consistent with levels found in an estuary.     

Single-cell Iso-Sequencing enables rapid genome annotation for scRNAseq analysis

Single-cell RNA sequencing is a powerful technique that continues to expand across various biological applications. However, incomplete 3′-UTR annotations can impede single-cell analysis resulting in genes that are partially or completely uncounted. Performing single-cell RNA sequencing with incomplete 3′-UTR annotations can hinder the identification of cell identities and gene expression patterns and lead to erroneous biological inferences. We demonstrate that performing single-cell isoform sequencing in tandem with single-cell RNA sequencing can rapidly improve 3′-UTR annotations. Using threespine stickleback fish (Gasterosteus aculeatus), we show that gene models resulting from a minimal embryonic single-cell isoform sequencing dataset retained 26.1% greater single-cell RNA sequencing reads than gene models from Ensembl alone. Furthermore, pooling our single-cell sequencing isoforms with a previously published adult bulk Iso-Seq dataset from stickleback, and merging the annotation with the Ensembl gene models, resulted in a marginal improvement (+0.8%) over the single-cell isoform sequencing only dataset. In addition, isoforms identified by single-cell isoform sequencing included thousands of new splicing variants. The improved gene models obtained using single-cell isoform sequencing led to successful identification of cell types and increased the reads identified of many genes in our single-cell RNA sequencing stickleback dataset. Our work illuminates single-cell isoform sequencing as a cost-effective and efficient mechanism to rapidly annotate genomes for single-cell RNA sequencing.     

Are We Predicting the Actual or Apparent Distribution of Temperate Marine Fishes?

Planning for resilience is the focus of many marine conservation programs and initiatives. These efforts aim to inform conservation strategies for marine regions to ensure they have inbuilt capacity to retain biological diversity and ecological function in the face of global environmental change--particularly changes in climate and resource exploitation. In the absence of direct biological and ecological information for many marine species, scientists are increasingly using spatially-explicit, predictive-modeling approaches. Through the improved access to multibeam sonar and underwater video technology these models provide spatial predictions of the most suitable regions for an organism at resolutions previously not possible. However, sensible-looking, well-performing models can provide very different predictions of distribution depending on which occurrence dataset is used. To examine this, we construct species distribution models for nine temperate marine sedentary fishes for a 25.7 km(2) study region off the coast of southeastern Australia. We use generalized linear model (GLM), generalized additive model (GAM) and maximum entropy (MAXENT) to build models based on co-located occurrence datasets derived from two underwater video methods (i.e. baited and towed video) and fine-scale multibeam sonar based seafloor habitat variables. Overall, this study found that the choice of modeling approach did not considerably influence the prediction of distributions based on the same occurrence dataset. However, greater dissimilarity between model predictions was observed across the nine fish taxa when the two occurrence datasets were compared (relative to models based on the same dataset). Based on these results it is difficult to draw any general trends in regards to which video method provides more reliable occurrence datasets. Nonetheless, we suggest predictions reflecting the species apparent distribution (i.e. a combination of species distribution and the probability of detecting it). Consequently, we also encourage researchers and marine managers to carefully interpret model predictions.     

Multi-scale approach for predicting fish species distributions across coral reef seascapes

Two of the major limitations to effective management of coral reef ecosystems are a lack of information on the spatial distribution of marine species and a paucity of data on the interacting environmental variables that drive distributional patterns. Advances in marine remote sensing, together with the novel integration of landscape ecology and advanced niche modelling techniques provide an unprecedented opportunity to reliably model and map marine species distributions across many kilometres of coral reef ecosystems. We developed a multi-scale approach using three-dimensional seafloor morphology and across-shelf location to predict spatial distributions for five common Caribbean fish species. Seascape topography was quantified from high resolution bathymetry at five spatial scales (5–300 m radii) surrounding fish survey sites. Model performance and map accuracy was assessed for two high performing machine-learning algorithms: Boosted Regression Trees (BRT) and Maximum Entropy Species Distribution Modelling (MaxEnt). The three most important predictors were geographical location across the shelf, followed by a measure of topographic complexity. Predictor contribution differed among species, yet rarely changed across spatial scales. BRT provided ‘outstanding’ model predictions (AUC = >0.9) for three of five fish species. MaxEnt provided ‘outstanding’ model predictions for two of five species, with the remaining three models considered ‘excellent’ (AUC = 0.8–0.9). In contrast, MaxEnt spatial predictions were markedly more accurate (92% map accuracy) than BRT (68% map accuracy). We demonstrate that reliable spatial predictions for a range of key fish species can be achieved by modelling the interaction between the geographical location across the shelf and the topographic heterogeneity of seafloor structure. This multi-scale, analytic approach is an important new cost-effective tool to accurately delineate essential fish habitat and support conservation prioritization in marine protected area design, zoning in marine spatial planning, and ecosystem-based fisheries management.     

Shortcomings of current fish transfer functions and a proposal for a new transfer function

One important component of almost all theoretical models in fishery is a fish transfer function. However, most of the current fish transfer functions have significant shortcomings. This paper contributes to the literature on fishery management by (1) showing some of shortcomings of commonly used fish transfer functions and proposing a new fish transfer function that is more appropriate to model net amount of fish transfer from one marine area to another; and (2) applying the proposed transfer function in an optimal harvest problem to assess the economic payoff from a switching reserve versus a fixed marine reserve. The findings indicate that a switching marine reserve appears to provide fishers with higher economic benefits than a fixed marine reserve. The payoff gain from a switching reserve appears to increase when the fish move less because of bio-ecological and territorial factors that impede the fish dispersal between marine areas.     

MAD-YOLO: A quantitative detection algorithm for dense small-scale marine benthos

Marine biological resources are abundant, and the reasonable development, research and protection of marine biological resources are of great significance to marine ecological health and economic development. At present, underwater object quantitative detection plays a very important role in marine biological science research, marine species richness survey, and rare species conservation. However, the problems of a large amount of noise in the underwater environment, small object scale, dense biological distribution, and occlusion all increase the detection difficulty. In this paper, a detection algorithm MAD-YOLO (Multiscale Feature Extraction and Attention Feature Fusion Reinforced YOLO for Marine Benthos Detection) is proposed, which is based on improved YOLOv5 is proposed to solve the above problems. To improve the adaptability of the network to the underwater environment, VOVDarkNet is designed as the feature extraction backbone. It uses the intermediate features with different receptive fields to reinforce the ability to extract feature. AFC-PAN is proposed as the feature fusion network so that the network can learn correct feature information and location information of objects at various scales, improving the network's ability to perceive small objects. Label assignment strategy SimOTA and decoupled head are introduced to help the model better handles occlusion and dense distribution problems. Experiments show the MAD-YOLO algorithm increases mAP_0.5:0.95 on the URPC2020 dataset from 49.8% to 53.4% compared to the original YOLOv5. Moreover, the advantages of the model are visualized and analyzed by the method of controlling variables in the experimental part. The experiments show that MAD-YOLO is suitable for detecting blurred, dense, and small-scale objects. The model performs well in marine benthos detection tasks and can effectively promote marine life science research and marine engineering implementation. The source code is publicly available at https://github.com/JoeNan1/MAD-YOLO.     

DeepWILD: Wildlife Identification,Localisation and estimation on camera trap videos using Deep learning

Videos and images from camera traps are more and more used by ecologists to estimate the population of species on a territory. It is a laborious work since experts have to analyse massive data sets manually. This takes also a lot of time to filter these videos when many of them do not contain animals or are with human presence. Fortunately, deep learning algorithms for object detection can help ecologists to identify multiple relevant species on their data and to estimate their population. In this study, we propose to go even further by using object detection model to detect, classify and count species on camera traps videos. To this end, we developed a 3-step process: (i) At the first stage, after splitting videos into images, we annotate images by associating bounding boxes to each label thanks to MegaDetector algorithm; (ii) then, we extend MegaDetector based on Faster R-CNN architecture with backbone Inception-ResNet-v2 in order to not only detect the 13 relevant classes but also to classify them; (iii) finally, we design a method to count individuals based on the maximum number of bounding boxes detected. This final stage of counting is evaluated in two different contexts: first including only detection results (i.e. comparing our predictions against the right number of individuals, no matter their true class), then an evolved version including both detection and classification results (i.e. comparing our predictions against the right number in the right class). The results obtained during the evaluation of our model on the test data set are: (i) 73,92% mAP for classification, (ii) 96,88% mAP for detection with a ratio Intersection-Over-Union (IoU) of 0.5 (overlapping ratio between groundtruth bounding box and the detected one), and (iii) 89,24% mAP for detection at IoU = 0.75. Highly represented classes, like humans, have highest values of mAP around 81% whereas less represented classes in the train data set, such as dogs, have lowest values of mAP around 66%. Regarding the proposed counting method, we predicted a count either exact or $\pm$ 1 unit for 87% with detection results and for 48% with detection and classification results of our test data set. Our model is also able to detect empty videos. To the best of our knowledge, this is the first study in France about the use of object detection model on a French national park to locate, identify and estimate the population of species from camera trap videos.     

Strain variation in Mycobacterium marinum fish isolates

A molecular characterization of two Mycobacterium marinum genes, 16S rRNA and hsp65, was carried out with a total of 21 isolates from various species of fish from both marine and freshwater environments of Israel, Europe, and the Far East. The nucleotide sequences of both genes revealed that all M. marinum isolates from fish in Israel belonged to two different strains, one infecting marine (cultured and wild) fish and the other infecting freshwater (cultured) fish. A restriction enzyme map based on the nucleotide sequences of both genes confirmed the divergence of the Israeli marine isolates from the freshwater isolates and differentiated the Israeli isolates from the foreign isolates, with the exception of one of three Greek isolates from marine fish which was identical to the Israeli marine isolates. The second isolate from Greece exhibited a single base alteration in the 16S rRNA sequence, whereas the third isolate was most likely a new Mycobacterium species. Isolates from Denmark and Thailand shared high sequence homology to complete identity with reference strain ATCC 927. Combined analysis of the two gene sequences increased the detection of intraspecific variations and was thus of importance in studying the taxonomy and epidemiology of this aquatic pathogen. Whether the Israeli M. marinum strain infecting marine fish is endemic to the Red Sea and found extremely susceptible hosts in the exotic species imported for aquaculture or rather was accidentally introduced with occasional imports of fingerlings from the Mediterranean Sea could not be determined.     

The basis of fish community structure: marine and freshwater comparisons

Synopsis Freshwater and marine fish communities are described and compared for arctic, boreal and tropical latitudes. Details of habitat characteristics, species numbers, and diel and seasonal differences in distribution are given for each community type. The order of increasing richness of fish species in these environments is (1) arctic lakes, (2) arctic marine, (3) boreal lakes, (4) tropical lakes, (5) boreal marine and (6) tropical marine. The richness of numbers of species can be related to a series of factors, each of which may function at some threshold value. These factors include climatic perturbation, solar radiation, spatial heterogeneity, available nutrient supply, availability of cover, and geological time. Discontinuities in the availability of some factors can be partially compensated for by torpor or aestivation; this effectively removes the fish from the community for a period of time. Increased diversity may also be effected through the diurnal/nocturnal shift in activity in some fish communities.The development of an organic matrix, notably macrophyte beds or coral reefs, may contribute significantly to an increase in diversity within fish communities. This matrix operates by an increase in spatial heterogeneity and in biological interactions. The apparent lack of resilience of high diversity fish communities can be related to the characteristics of the underlying organic matrix. A change in the matrix will cause a change in the level of fish diversity that can be maintained in the system.This paper forms part of the proceedings of a mini-symposium convened at Cornell University, Ithaca, N.Y., 18–19 May 1976, entitled Patterns of Community Structure in Fishes (G.S. Helfman, ed.).     

Marine dock pilings foster diverse,native cryptobenthic fish assemblages across bioregions

Anthropogenic habitats are increasingly prevalent in coastal marine environments. Previous research on sessile epifauna suggests that artificial habitats act as a refuge for nonindigenous species, which results in highly homogenous communities across locations. However, vertebrate assemblages that live in association with artificial habitats are poorly understood. Here, we quantify the biodiversity of small, cryptic (henceforth “cryptobenthic”) fishes from marine dock pilings across six locations over 35° of latitude from Maine to Panama. We also compare assemblages from dock pilings to natural habitats in the two southernmost locations (Panama and Belize). Our results suggest that the biodiversity patterns of cryptobenthic fishes from dock pilings follow a Latitudinal Diversity Gradient (LDG), with average local and regional diversity declining sharply with increasing latitude. Furthermore, a strong correlation between community composition and spatial distance suggests distinct regional assemblages of cryptobenthic fishes. Cryptobenthic fish assemblages from dock pilings in Belize and Panama were less diverse and had lower densities than nearby reef habitats. However, dock pilings harbored almost exclusively native species, including two species of conservation concern absent from nearby natural habitats. Our results suggest that, in contrast to sessile epifaunal assemblages on artificial substrates, artificial marine habitats can harbor diverse, regionally characteristic assemblages of vertebrates that follow macroecological patterns that are well documented for natural habitats. We therefore posit that, although dock pilings cannot function as a replacement for natural habitats, dock pilings may provide cost‐effective means to preserve native vertebrate biodiversity, and provide a habitat that can be relatively easily monitored to track the status and trends of fish biodiversity in highly urbanized coastal marine environments.     

Importance of Photobacterium phosphoreum in relation to spoilage of modified atmosphere-packed fish products

Occurrence and growth of Photobacterium phosphoreum were studied in 20 experiments with fresh fish from Denmark, Iceland and Greece. The organism was detected in all marine fish species but not in fish from fresh water. Growth of P. phosphoreum to high levels (>10⁷ cfu g⁻¹) was observed in most products and the organism is likely to be of importance for spoilage of several modified atmosphere-packed (MAP) marine fish species when stored at chill temperatures. Some microbiological methods recommended for control of fish products by national and international authorities are inappropriate for detection of psychrotolerant and heat-labile micro-organisms like P. phosphoreum . These methods have been used in many previous studies of MAP fish and this could explain why, contrary to the findings in the present study, P. phosphoreum in general was not detected previously in spoiled MAP fish.     

Automatic stomata recognition and measurement based on improved YOLO deep learning model and entropy rate superpixel algorithm

The traditional methods of analyzing stomatal traits are mostly manual observation and measurement. These methods are time-consuming, labor-intensive, and inefficient. Some methods have been proposed for the automatic recognition and counting of stomata, however most of those methods could not complete the automatic measurement of stomata parameters at the same time. Some non-deep learning methods could automatically measure the parameters of stomata, but they could not complete the automatic recognition and detection of stomata. In this paper, a deep learning-based method was proposed for automatically identifying, counting and measuring stomata of maize (Zea mays L.) leaves at the same time. An improved YOLO (You Only Look Once) deep learning model was proposed to identify stomata of maize leaves automatically, and an entropy rate superpixel algorithm was used for the accurate measurement of stomatal parameters. According to the characteristics of the stomata images data set, the network structure of YOLOv5 was modified, which greatly reduced the training time without affecting the recognition performance. The predictor in YOLO deep learning model was optimized, which reduced the false detection rate. At the same time, the 16-fold and 32-fold down-sampling layers were simplified according to the characteristics of stomatal objects, which improved the recognition efficiency. Experimental results showed that the recognition precision of the improved YOLO deep learning model reached 95.3% on the maize leaves stomatal data set, and the average accuracy of parameter measurement reached 90%. The proposed method could fully automatically complete the recognition, counting and measurement of stomata of plants, which can help agricultural scientists and botanists to conduct large-scale researches of stomatal morphology, structure and physiology, as well as the researches combined with genetic analysis or molecular-level analysis.