Biodiversity Information Science and Standards最新文献

{"title":"Next Steps Towards Better Living Atlas Deployments and Maintenance","authors":"David Martin, Vicente Ruiz Jurado","doi":"10.3897/biss.7.112560","DOIUrl":"https://doi.org/10.3897/biss.7.112560","url":null,"abstract":"The Living Atlases project, facilitated by the Global Biodiversity Information Facility (GBIF) and the Atlas of Living Australia (ALA), has successfully operated for more than a decade, establishing collaborations with over 30 countries who have implemented ALA components within their respective environments. Over this period, technological advancements and the prevalence of cloud platforms have transformed the landscape of infrastructure management. In this presentation, we will explore innovative approaches to streamline the installation process of ALA, capitalizing on the benefits offered by cloud platforms and cutting-edge technologies. Furthermore, ALA has maintained a strong collaborative partnership with GBIF over the past four years, focusing on data ingestion pipelines and, more recently, engaging in shared user interface (UI) development. These improvements aim to enhance the maintainability of ALA modules, enabling organizations to leverage the advantages provided by cloud-based solutions and novel technologies.","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135878450","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Demonstration of Taxonomic Name Data Services through ChecklistBank","authors":"Olaf Bánki, Markus Döring, Thomas Jeppesen, Donald Hobern","doi":"10.3897/biss.7.112544","DOIUrl":"https://doi.org/10.3897/biss.7.112544","url":null,"abstract":"ChecklistBank is a publishing platform and open data repository focused on taxonomic and nomenclatural datasets. It was launched at the end of 2020, and is a joint development by Catalogue of Life (COL) and the Global Biodiversity Information Facility (GBIF). Close to 50K datasets, mostly originating from published literature mediated through Plazi's TreatmentBank, Pensoft Publishers and the European Journal of Taxonomy, are openly accessible through ChecklistBank. Data sets also include sources with (Molecular) Operational Taxonomic Units, such as from UNITE / PlutoF, National Center for Biotechnology Information Taxonomy / European Nuclotide Archive, and the International Barcode of Life / BoLD. Next to various taxonomic datasets (also from regional and national levels, e.g., shared through GBIF) and nomenclatural datasets (e.g., Zoobank, International Plant Names Index), ChecklistBank also links out to the various original initiatives websites (e.g., World Register of Marine Species, Integrated Taxonomic Information System, COL China, Species Files). ChecklistBank also holds all the tooling that is needed to assemble the COL Checklist, the authoritative global species list of all described organisms. The COL Checklist 2023 (Bánki et al. 2023), containing more than 2.1 million accepted species, is assembled from 164 global taxonomic data sources mediated through ChecklistBank. The COL Checklist contains name usage identifiers, and each checklist version and its underpinning data sources are issued with digital object identifiers. After the launching of ChecklistBank, the EU funded Biodiversity Community Integrated Knowledge Library (BiCIKL) project contributed to additional improvements to ChecklistBank. These added functionalities include, amongst others, a name usage search, name match, and a taxonomic data comparison. The tooling used to assemble the COL Checklist has been generalised through a ChecklistBank 'project functionality' supporting the assembly of a species list. During the demonstration, several of the functionalities, developed in the context of the EU BiCIKL project, will be highlighted.","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135831397","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Synergizing Digital, Biological, and Participatory Sciences for Global Plant Species Identification: Enabling access to a worldwide identification service","authors":"Pierre Bonnet, Antoine Affouard, Jean-Christophe Lombardo, Mathias Chouet, Hugo Gresse, Vanessa Hequet, Remi Palard, Maxime Fromholtz, Vincent Espitalier, Hervé Goëau, Benjamin Deneu, Christophe Botella, Joaquim Estopinan, César Leblanc, Maximilien Servajean, François Munoz, Alexis Joly","doi":"10.3897/biss.7.112545","DOIUrl":"https://doi.org/10.3897/biss.7.112545","url":null,"abstract":"Human activities have a growing impact on global biodiversity. While our understanding of biodiversity worldwide is not yet comprehensive, it is crucial to explore effective means of characterizing it in order to mitigate these impacts. The advancements in data storage, exchange capabilities, and the increasing availability of extensive taxonomic, ecological, and environmental databases offer possibilities for implementing new approaches that can address knowledge gaps regarding species and habitats. This enhanced knowledge will, in turn, facilitate improved management practices and enable better local governance of territories. Meeting these requirements necessitates the development of innovative tools and methods to respond to these needs. Citizen science platforms have emerged as valuable resources for generating large amounts of biodiversity data, thanks to their visibility and attractiveness to individuals involved in territorial management and education. These platforms present new opportunities to train deep learning models for automated species recognition, leveraging the substantial volumes of multimedia data they accumulate. However, effectively managing, curating, and disseminating the data and services generated by these platforms remains a significant challenge that hinders the achievement of their objectives. In line with this, the GUARDEN and MAMBO European projects aim to utilize the Pl@ntNet participatory science platform (Affouard et al. 2021) to develop and implement novel computational services to enable the widespread creation of floristic inventories. In the pursuit of this project, various standards and reference datasets have been employed, such as the POWO (Plants of the World Online) world checklist and the WGSRPD (World Geographical Scheme for Recording Plant Distributions) standard, to establish a foundation for creating a global service that aids in plant identification through visual analysis. This service relies on a NoSQL (Not Only Structured Query Language) data management system ArangoDB (Arango Database), utilizes state-of-the-art automated visual classification models (vision transformers), and operates on a distributed IT (Information Technology) infrastructure that leverages the capabilities of collaborative stakeholders interested in supporting this initiative. Global-scale automated workflows have been established specifically for the collection, analysis, and dissemination of illustrated occurrences of plant species. These workflows now enable the development of new IT tools that facilitate the description and monitoring of species and habitat conservation statuses. A comprehensive presentation highlighting the significant advancements achieved will be provided to share the lessons learned during its development and ensure the widespread adoption of this service within the scientific community.","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135878792","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Recognising Indigenous Provenance in Biodiversity Records","authors":"Maui Hudson, Jane Anderson, Riley Taitingfong, Andrew Martinez, Stephanie Carroll","doi":"10.3897/biss.7.112610","DOIUrl":"https://doi.org/10.3897/biss.7.112610","url":null,"abstract":"The advent of data-driven technologies and the increasing demand for data have brought about unique challenges for Indigenous data governance. The CARE principles emphasize Collective Benefit, Authority, Responsibility, and Ethics as essential pillars for ensuring that Indigenous data rights are upheld, Indigenous knowledge is protected, and Indigenous Peoples are active participants in data governance processes (Carroll et al. 2020, Carroll et al. 2021). Identifying tangible activities and providing guidance to centre Indigenous perspectives provide a comprehensive approach to address the complexities of Indigenous data governance in a rapidly evolving data landscape (Gupta et al. 2023, Jennings et al. 2023, Sterner and Elliott 2023). Biodiversity research has increasingly recognized the intertwined relationship between biological diversity and cultural practices, leading to discussions about how research can strengthen the evidence base, build trust, enhance legitimacy for decision making (Alexander et al. 2021) and explore requirements for Indigenous metadata (Jennings et al. 2023). An Indigenous Metadata Bundle Communique, produced following an Indigenous Metadata Symposium, recommended the initial categories as: Governance, Provenance, Lands & Waters, Protocols, and Local Contexts Notices & Labels. Traditional Knowledge (TK) and Biocultural (BC) Labels have emerged as essential tools for recognising and maintaining Indigenous provenance, protocols and permissions in records for both natural ecosystems and cultural heritage (Anderson et al. 2020, Liggins et al. 2021) emphasizing the importance of Indigenous Peoples and local knowledge systems in research and digital management. Biocultural labels acknowledge the intricate links between biodiversity and cultural diversity, emphasizing the role of indigenous communities in preserving biodiversity through their traditional practices (Hudson et al. 2021). By recognizing the intrinsic value of these relationships, TK and BC labels not only contribute to a more holistic understanding of biodiversity but also promote ethical considerations and mutual respect between researchers and local communities, fostering collaborative partnerships for research and conservation initiatives (McCartney et al. 2023). Addressing the CARE Principles for Indigenous Data Governance in biodiversity research introduces several challenges and opportunities. Ethical concerns regarding recognition of Indigenous rights and interests in data (Hudson et al. 2023), intellectual property rights, cultural appropriation, and equitable benefit sharing, must be navigated sensitively (Carroll et al. 2022b, Golan et al. 2022). Moreover, fostering effective communication between researchers and communities is paramount for ensuring the accuracy and authenticity of Indigenous metadata and protocols for appropriate use (Carroll et al. 2022a). However, these challenges are offset by the potential for enriching scientific knowledg","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135831122","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Implementing the CARE Principles for Indigenous Data Governance in Biodiversity Data Management","authors":"Riley Taitingfong, Stephanie Carroll","doi":"10.3897/biss.7.112615","DOIUrl":"https://doi.org/10.3897/biss.7.112615","url":null,"abstract":"Indigenous data governance is a critical aspect of upholding Indigenous rights and fostering equitable partnerships in biodiversity research and data management. An estimated 80% of the planet’s biodiversity exists on Indigenous lands (Sobrevila 2008), and the majority of Indigenous data derived from specimens taken from Indigenous lands are held by non-Indigenous entities and institutions. The CARE Principles (Collective benefit, Authority to control, Responsibility, and Ethics) are designed to guide the inclusion of Indigenous peoples in data governance, and increase their access to and benefit from data (Carroll et al. 2020). This talk will share emerging tools and resources that can be leveraged to implement the CARE Principles within repositories and institutions that hold Indigenous data. This talk highlights two primary tools to promote Indigenous data governance in repositories: a phased framework to guide third-party holders of Indigenous data through foundational learning and concrete steps to apply the CARE principles in their respective settings, and the CARE criteria, an assessment tool by which researchers and institutions can evaluate the maturity of CARE implementation, identify areas for improvement, and allow other entities such as funders and publishers to evaluate CARE compliance. a phased framework to guide third-party holders of Indigenous data through foundational learning and concrete steps to apply the CARE principles in their respective settings, and the CARE criteria, an assessment tool by which researchers and institutions can evaluate the maturity of CARE implementation, identify areas for improvement, and allow other entities such as funders and publishers to evaluate CARE compliance.","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135830419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The Role of the CLIP Model in Analysing Herbarium Specimen Images","authors":"Vamsi Krishna Kommineni, Jens Kattge, Jitendra Gaikwad, Susanne Tautenhahn, Birgitta Koenig-ries","doi":"10.3897/biss.7.112566","DOIUrl":"https://doi.org/10.3897/biss.7.112566","url":null,"abstract":"The number of openly-accessible digital plant specimen images is growing tremendously and available through data aggregators: Global Biodiversity Information Facility (GBIF) contains 43.2 million images, and Intergrated Digitized Biocollections (iDigBio) contains 32.4 million images (Accessed on 29.06.2023). All these images contain great ecological (morphological, phenological, taxonomic etc.) information, which has the potential to facilitate the conduct of large-scale analyses. However, extracting this information from these images and making it available to analysis tools remains challenging and requires more advanced computer vision algorithms. With the latest advancements in the natural language processing field, it is becoming possible to analyse images with text prompts. For example, with the Contrastive Language-Image Pre-Training (CLIP) model, which was trained on 400 million image-text pairs, it is feasible to classify day-to-day life images by providing different text prompts and an image as an input to the model, then the model can predict the most suitable text prompt for the input image. We explored the feasibility of using the CLIP model to analyse digital plant specimen images. A particular focus of this study was on the generation of appropriate text prompts. This is important as the prompt has a large influence on the results of the model. We experimented with three different methods: a) automatic text prompt based on metadata of the specific image or other datasets, b) automatic generic text prompt of the image (describing what is in the image) and c) manual text prompt by annotating the image. We investigated the suitability of these prompts with an experiment, where we tested whether the CLIP model could recognize a herbarium specimen image using digital plant specimen images and semantically disparate text prompts. Our ultimate goal is to filter the digital plant specimen images based on the availability of intact leaves and measurement scale to reduce the number of specimens that reach the downstream pipeline, for instance, the segmentation task for the leaf trait extraction process. To achieve the goal, we are fine-tuning the CLIP model with a dataset of around 20,000 digital plant specimen image-text prompt pairs, where the text prompts were generated using different datasets, metadata and generic text prompt methods. Since the text prompts can be created automatically, it is possible to eradicate the laborious manual annotating process. In conclusion, we present our experimental testing of the CLIP model on digital plant specimen images with varied settings and how the CLIP model can act as a potential filtering tool. In future, we plan to investigate the possibility of using text prompts to do the instance segmentation to extract leaf trait information using Large Language Models (LLMs).","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135879248","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Community Curation of Nomenclatural and Taxonomic Information in the Context of the Collection Management System JACQ","authors":"Heimo Rainer, Andreas Berger, Tanja Schuster, Johannes Walter, Dieter Reich, Kurt Zernig, Jiří Danihelka, Hana Galušková, Patrik Mráz, Natalia Tkach, Jörn Hentschel, Jochen Müller, Sarah Wagner, Walter Berendsohn, Robert Lücking, Robert Vogt, Lia Pignotti, Francesco Roma-Marzio, Lorenzo Peruzzi","doi":"10.3897/biss.7.112571","DOIUrl":"https://doi.org/10.3897/biss.7.112571","url":null,"abstract":"Nomenclatural and taxonomic information are crucial for curating botanical collections. In the course of changing methods for systematic and taxonomic studies, classification systems changed considerably over time (Dalla Torre and Harms 1900, Durand and Bentham 1888, Endlicher 1836, Angiosperm Phylogeny Group et al. 2016). Various approaches to store preserved material have been implemented, most of them based on scientific names (e.g., families, genera, species) often in combination with other criteria such as geographic provenance or collectors. The collection management system, JACQ, was established in the early 2000s then developed to support multiple institutions. It features a centralised data storage (with mirror sites) and access via the Internet. Participating collections can download their data at any time in a comma-separated values (CSV) format. From the beginning, JACQ was conceived as a collaboration platform for objects housed in botanical collections, i.e., plant, fungal and algal groups. For these groups, various sources of taxonomic reference exist, nowadays online resources are preferred, e.g., Catalogue of Life, AlgaeBase, Index Fungorum, Mycobank, Tropicos, Plants of the World Online, International Plant Names Index (IPNI), World Flora Online, Euro+Med, Anthos, Flora of Northamerica, REFLORA, Flora of China, Flora of Cuba, Australian Virtual Herbarium (AVH). Implementation and (re)use of PIDs Persistent identifiers (PIDs) for names (at any taxonomic rank) apart from PIDs for taxa, are essential to allow and support reliable referencing across institutions and thematic research networks (Agosti et al. 2022). For this purpose we have integrated referencing to several of the above mentioned resources and populate the names used inside JACQ with those external PIDs. For example, Salix rosmarinifolia is accepted in Plants of the World Online while Euro+Med Plantbase considers it a synonym of Salix repens subsp. rosmarinifolia. Either one can be an identification of a specimen in the JACQ database. Retrieval of collection material One strong use case is the curation of material in historic collections. On the basis of outdated taxon concepts that were applied to the material in history, \"old\" synonyms are omnipresent in historical collections. In order to retrieve all material of a given taxon, it is necessary to know all relevant names. Future outlook In combination with the capability of Linked Data and the IIIF (International Image Interoperability Framework) technology, these PIDs serve as crucial elements for the integration of decentralized information systems and reuse of (global) taxonomic backbones in combination with collection management systems (Gamer and Kreyenbühl 2022, Hyam 2022, Loh 2017).","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135879283","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Extracting Reproductive Condition and Habitat Information from Text Using a Transformer-based Information Extraction Pipeline","authors":"Roselyn Gabud, Nelson Pampolina, Vladimir Mariano, Riza Batista-Navarro","doi":"10.3897/biss.7.112505","DOIUrl":"https://doi.org/10.3897/biss.7.112505","url":null,"abstract":"Understanding the biology underpinning the natural regeneration of plant species in order to make plans for effective reforestation is a complex task. This can be aided by providing access to databases that contain long-term and wide-scale geographical information on species distribution, habitat, and reproduction. Although there exists widely-used biodiversity databases that contain structured information on species and their occurrences, such as the Global Biodiversity Information Facility (GBIF) and the Atlas of Living Australia (ALA), the bulk of knowledge about biodiversity still remains embedded in textual documents. Unstructured information can be made more accessible and useful for large-scale studies if there are tools and services that automatically extract meaningful information from text and store it in structured formats, e.g., open biodiversity databases, ready to be consumed for analysis (Thessen et al. 2022). We aim to enrich biodiversity occurrence databases with information on species reproductive condition and habitat, derived from text. In previous work, we developed unsupervised approaches to extract related habitats and their locations, and related reproductive condition and temporal expressions (Gabud and Batista-Navarro 2018). We built a new unsupervised hybrid approach for relation extraction (RE), which is a combination of classical rule-based pattern-matching methods and transformer-based language models that framed our RE task as a natural language inference (NLI) task. Using our hybrid approach for RE, we were able to extract related biodiversity entities from text even without a large training dataset. In this work, we implement an information extraction (IE) pipeline comprised of a named entity recognition (NER) tool and our hybrid relation extraction (RE) tool. The NER tool is a transformer-based language model that was pretrained on scientific text and then fine-tuned using COPIOUS (Conserving Philippine Biodiversity by Understanding big data; Nguyen et al. 2019), a gold standard corpus containing named entities relevant to species occurrence. We applied the NER tool to automatically annotate geographical location, temporal expression and habitat information contained within sentences. A dictionary-based approach is then used to identify mentions of reproductive conditions in text (e.g., phrases such as \"fruited heavily\" and \"mass flowering\"). We then use our hybrid RE tool to extract reproductive condition - temporal expression and habitat - geographical location entity pairs. We test our IE pipeline on the forestry compendium available in the CABI Digital Library (Centre for Agricultural and Biosciences International), and show that our work enables the enrichment of descriptive information on reproductive and habitat conditions of species. This work is a step towards enhancing a biodiversity database with the inclusion of habitat and reproductive condition information extracted from text.","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135980580","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Application of AI-Helped Image Classification of Fish Images: An iDigBio dataset example","authors":"Bahadir Altintas, Yasin Bakış, Xiojun Wang, Henry Bart","doi":"10.3897/biss.7.112438","DOIUrl":"https://doi.org/10.3897/biss.7.112438","url":null,"abstract":"Artificial Intelligence (AI) becomes more prevalent in data science as well as in areas of computational science. Commonly used classification methods in AI can also be used for unorganized databases, if a proper model is trained. Most of the classification work is done on image data for purposes such as object detection and face recognition. If an object is well detected from an image, the classification may be done to organize image data. In this work, we try to identify images from an Integrated Digitized Biocollections (iDigBio) dataset and to classify these images to generate metadata to use as an AI-ready dataset in the future. The main problem of the museum image datasets is the lack of metadata information on images, wrong categorization, or poor image quality. By using AI, it maybe possible to overcome these problems. Automatic tools can help find, eliminate or fix these problems. For our example, we trained a model for 10 classes (e.g., complete fish, photograph, notes/labels, X-ray, CT (computerized tomotography) scan, partial fish, fossil, skeleton) by using a manually tagged iDigBio image dataset. After training a model for each for class, we reclassified the dataset by using these trained models. Some of the results are given in Table 1. As can be seen in the table, even manually classified images can be identified as different classes, and some classes are very similar to each other visually such as CT scans and X-rays or fossils and skeletons. Those kind of similarities are very confusing for the human eye as well as AI results.","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135980908","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An AI-based Wild Animal Detection System and Its Application","authors":"Congtian Lin, Jiangning Wang, Liqiang Ji","doi":"10.3897/biss.7.112456","DOIUrl":"https://doi.org/10.3897/biss.7.112456","url":null,"abstract":"Rapid accumulation of biodiversity data and development of deep learning methods bring the opportunities for detecting and identifying wild animals automatically, based on artificial intelligence. In this paper, we introduce an AI-based wild animal detection system. It is composed of acoustic and image sensors, network infrastructures, species recognition models, and data storage and visualization platform, which go through the technical chain learned from Internet of Things (IOT) and applied to biodiversity detection. The workflow of the system is as follows: Deploying sensors for different detection targets . The acoustic sensor is composed of two microphones for picking up sounds from the environment and an edge computing box for judging and sending back the sound files. The acoustic sensor is suitable for monitoring birds, mammals, chirping insects and frogs. The image sensor is composed of a high performance camera that can be controlled to record surroundings automatically and a video analysis edge box running a model for detecting and recording animals. The image sensor is suitable for monitoring waterbirds in locations without visual obstructions. Adopting different networks according to signal availability . Network infrastructures are critical for the detection system and the task of transferring data collected by sensors. We use the existing network when 4/5G signals are available, and build special networks using Mesh Networking technology for the areas without signals. Multiple network strategies lower the cost for monitoring jobs. Recognizing species from sounds, images or videos . AI plays a key role in our system. We have trained acoustic models for more than 800 Chinese birds and some common chirping insects and frogs, which can be identified from sound files recorded by acoustic sensors. For video and image data, we also have trained models for recognizing 1300 Chinese birds and 400 mammals, which help to discover and count animals captured by image sensors. Moreover, we propose a special method for detecting species through features of voices, images and niche features of animals. It is a flexible framework to adapt to different combinations of acoustic and image sensors. All models were trained with labeled voices, images and distribution data from Chinese species database, ESPECIES. Saving and displaying machine observations . The original sound, image and video files with identified results were stored in the data platform deployed on the cloud for extensible computing and storage. We have developed visualization modules in the platform for displaying sensors on maps using WebGIS to show curves of the number of records and species for each day, real time alerts from sensors capturing animals, and other parameters. Deploying sensors for different detection targets . The acoustic sensor is composed of two microphones for picking up sounds from the environment and an edge computing box for judging and sending back the sound fil","PeriodicalId":9011,"journal":{"name":"Biodiversity Information Science and Standards","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135981784","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}