Dataset for "Future of Edge AI in biodiversity monitoring". Review Analysis of 82 Publications 2017-2025

Dataset Description This dataset accompanies the systematic review "Future of Edge AI in biodiversity monitoring" and comprises structured metadata extracted from 82 publications (2017–2025) describing edge computing and edge AI systems for biodiversity monitoring. The .xlsx file contains one row per reviewed study and records bibliographic information (title, authors, year, DOI), classification variables (edge AI architecture type, sensing modality, monitored taxa, ecological application), and detailed technical specifications spanning hardware platforms and processing units, AI models and optimisation techniques, wireless connectivity and data transfer strategies, power source and energy consumption, deployment scale, duration, and geographic location, and estimated hardware cost where reported. The dataset supports the quantitative synthesis presented in the review and enables reproduction of all summary figures and statistics reported in the manuscript. Missing data: Cells with N/A indicate that the corresponding information was not reported in the source publication and was not inferable from associated repositories or supplementary materials. Variables Bibliographic and study information Variable Description Ref. # Reference identifier assigned to the study in the dataset. Year Publication year of the study. Title Title of the publication. Authors Authors of the publication. DOI/URL DOI or URL identifying the publication, where available. Abstract Abstract text recorded from the source publication. GitHub/Other Information on GitHub repositories or other external resources associated with the study. Type Edge-system type or system category assigned to the study. Edge AI, connectivity and sensing Variable Description EdgeAI Indicates whether AI processing is performed at the edge/on-device. Connectivity Indicates whether network or communication connectivity is reported. Sensing_Modality Primary sensing modality used for biodiversity or environmental monitoring. Deployment Context in which the system was deployed or evaluated. Taxa Broad taxonomic group monitored by the system. Taxa_Detailed More specific taxonomic information, including species or other detailed classifications where reported. Hardware and AI Variable Description HW_Platform Hardware platform(s) used for sensing, processing, or edge-AI computation. HW_Platform_Clean Standardised hardware-platform category used for comparison across studies. Main_Processing_Unit Main processor or computational unit used for processing and/or AI inference. AI-Model AI or machine-learning model/algorithm used by the system. AI_Optimisations AI/model optimisation techniques reported, such as quantisation, pruning, compression, or other approaches for edge deployment. AI_Inference_Framework Software framework, library, runtime, or inference engine used to execute the AI model. Connectivity_Type Communication technology reported for the system. Power_Source Primary power source reported for the system. Ecological application and sensing Variable Description Sensing_Specifications Technical specifications of the sensing configuration reported in the study. HW_Platform_Details Detailed description of the hardware platform and associated components. HW_Sensors Sensors used to collect biodiversity or environmental data. HW_Connectivity Hardware components or modules used to provide connectivity. Pre-Processing Data-processing steps applied before AI inference or subsequent analysis. AI-Model_Details Additional details on the AI model, architecture, implementation, or configuration. AI model and performance information Variable Description AI_Model_Metrics Performance metrics reported for the AI model or monitoring task. AI-Model_Size Reported size of the AI model, such as model parameters or memory/storage requirements. InferenceTime Reported time required to perform AI inference on the target hardware. DataTransferTime Reported time required to transfer data between system components or to a remote endpoint. Model Inference (Peak RAM Usage) Peak RAM usage reported during AI model inference. Model-ROMSpace ROM, flash, or other persistent storage space required by the AI model. Communication and data handling Variable Description Communication Specifications Technical specifications of the communication system, where reported. Data_Transfer_Details Specific types of data transmitted by the system, including raw observations, model outputs, summaries, or sensor readings. Data_Storage Data-storage configuration used by the monitoring system. Uplink Strategy Strategy used to transmit information from the monitoring device to a gateway, server, or other remote infrastructure. Power and deployment Variable Description Power Autonomy Reported operating duration or power autonomy of the system. Power_Source_Details Detailed description of the power supply and power-management configuration. Energy Consumption Reported energy or power consumption while the device is operating. Duty Cycle Reported duty cycle, sampling rate, sleep configuration, and/or wake-up policy. Deployment Scale Number of monitoring devices or nodes deployed, where reported. Deployment Duration Duration of the reported deployment. Location and cost Variable Description Location Geographic location or field site where the system was deployed or evaluated. Country, Continent Country and continent associated with the reported deployment location. Estimated Cost Reported or estimated cost of the system, device, or relevant hardware. Relevant Notes Additional information, assumptions, qualifications, or comments relevant to interpreting the extracted data.