Emerging technologies in citizen science and potential for insect monitoring

Philosophical Transactions of the Royal Society B Biological Sciences, Journal Year: 2024, Volume and Issue: 379(1904)

Published: May 5, 2024

Emerging technologies are increasingly employed in environmental citizen science projects. This integration offers benefits and opportunities for scientists participants alike. Citizen can support large-scale, long-term monitoring of species occurrences, behaviour interactions. At the same time, foster participant engagement, regardless pre-existing taxonomic expertise or experience, permit new types data to be collected. Yet, may also create challenges by potentially increasing financial costs, necessitating technological demanding training participants. Technology could reduce people's direct involvement engagement with nature. In this perspective, we discuss how current have spurred an increase projects implementation emerging enhance scientific impact public engagement. We show technology act as (i) a facilitator efforts, (ii) enabler research opportunities, (iii) transformer science, policy participation, but become (iv) inhibitor equity rigour. is developing fast promises provide many exciting insect monitoring, while seize these must remain vigilant against potential risks. article part theme issue ‘Towards toolkit global biodiversity monitoring’.

Language: Английский

---

Towards a toolkit for global insect biodiversity monitoring

Philosophical Transactions of the Royal Society B Biological Sciences, Journal Year: 2024, Volume and Issue: 379(1904)

Published: May 5, 2024

Insects are the most diverse group of animals on Earth, yet our knowledge their diversity, ecology and population trends remains abysmally poor. Four major technological approaches coming to fruition for use in insect monitoring ecological research—molecular methods, computer vision, autonomous acoustic radar-based remote sensing—each which has seen advances over past years. Together, they have potential revolutionize ecology, make all-taxa, fine-grained feasible across globe. So far, within among technologies largely taken place isolation, parallel efforts projects led redundancy a methodological sprawl; yet, given commonalities goals approaches, increased collaboration integration could provide unprecedented improvements taxonomic spatio-temporal resolution coverage. This theme issue showcases recent developments state-of-the-art applications these technologies, outlines way forward regarding data processing, cost-effectiveness, meaningful trend analysis, open requirements. papers set stage future automated monitoring. article is part ‘Towards toolkit global biodiversity monitoring’.

Language: Английский

---

Emerging technologies for pollinator monitoring

Current Opinion in Insect Science, Journal Year: 2025, Volume and Issue: unknown, P. 101367 - 101367

Published: March 1, 2025

Language: Английский

---

Delivering on a promise: futureproofing automated insect monitoring methods

Philosophical Transactions of the Royal Society B Biological Sciences, Journal Year: 2024, Volume and Issue: 379(1904)

Published: May 5, 2024

Due to rapid technological innovations, the automated monitoring of insect assemblages comes within reach. However, this continuous innovation endangers methodological continuity needed for calculating reliable biodiversity trends in future. Maintaining over prolonged periods time is not trivial, since technology improves, reference libraries grow and both hard- software used now may no longer be available Moreover, because data on many species are collected at same time, there will simple way calibrating outputs old new devices. To ensure that long-term can calculated using data, I make four recommendations: (1) Construct devices last decades, have a five-year overlap period when replaced. (2) resemble ones, especially some kind attractant (e.g. light) used. Keep extremely detailed metadata collection, detection identification methods, including attractants, enable this. (3) Store raw (sounds, images, DNA extracts, radar/lidar detections) future reprocessing with updated classification systems. (4) Enable forward backward compatibility processed example by in-silico 'degradation' match older quality. This article part theme issue 'Towards toolkit global monitoring'.

Language: Английский

---

Towards edge processing of images from insect camera traps

Remote Sensing in Ecology and Conservation, Journal Year: 2025, Volume and Issue: unknown

Published: April 17, 2025

Abstract Insects represent nearly half of all known multicellular species, but knowledge about them lags behind for most vertebrate species. In part this reason, they are often neglected in biodiversity conservation policies and practice. Computer vision tools, such as insect camera traps, automated monitoring have the potential to revolutionize study conservation. To further advance trapping analysis their image data, effective processing pipelines needed. paper, we present a flexible fast pipeline designed analyse these recordings by detecting, tracking classifying nocturnal insects broad taxonomy 15 classes resolution individual moth A classifier with anomaly detection is proposed filter dark, blurred or partially visible that will be uncertain classify correctly. simple track‐by‐detection algorithm track classified incorporating feature embeddings, distance area cost. We evaluated computational speed power performance different edge computing devices (Raspberry Pi's NVIDIA Jetson Nano) compared various time‐lapse (TL) strategies tracking. The minimum difference detections was found 2‐min TL intervals 0.5 frames per second; however, fewer than one night, Pearson correlation decreases. Shifting from would reduce number recorded images allow real‐time on trap Raspberry Pi. Nano energy‐efficient solution, capable at fps. Our applied more 5.7 million second 12 light traps during two full seasons located diverse habitats, including bogs, heaths forests. results thus show scalability traps.

Language: Английский

---

Insect Identification in the Wild: The AMI Dataset

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 55 - 73

Published: Dec. 1, 2024

Language: Английский

---

Towards edge processing of images from insect camera traps

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: July 3, 2024

Abstract Insects represent nearly half of all known organisms, with nocturnal insects being particularly challenging to monitor. Computer vision tools for automated monitoring have the potential revolutionize insect study and conservation. The advancement light traps camera-based systems necessitates effective flexible pipelines analysing recorded images. In this paper, we present a fast processing pipeline designed analyse these recordings by detecting, tracking classifying at taxonomic ranks order, suborder as well resolution individual moth species. consists four adaptable steps. first step detect in camera trap An order classifier anomaly detection is proposed filter dark, blurry or partly visible that will be uncertain classify correctly. A simple track-by-detection algorithm track classified incorporating feature embeddings, distance area cost. We evaluated computational speed power performance different edge computing devices (Raspberry Pi’s NVIDIA Jetson Nano) compared various time-lapse strategies tracking. minimum difference was found 2-minute intervals 0.5 frames per second, however, fewer than one night, Pearson correlation decreases. Shifting from would reduce amount images able perform real-time on Raspberry Pi. Nano most energy-efficient solution, capable fps. Our applied more 3.4 million second 12 during full season located diverse habitats, including bogs, heaths forests.

Language: Английский

---

InsectNet: Real-time identification of insects using an end-to-end machine learning pipeline

PNAS Nexus, Journal Year: 2024, Volume and Issue: 4(1)

Published: Dec. 23, 2024

Abstract Insect pests significantly impact global agricultural productivity and crop quality. Effective integrated pest management strategies require the identification of insects, including beneficial harmful insects. Automated insects under real-world conditions presents several challenges, need to handle intraspecies dissimilarity interspecies similarity, life-cycle stages, camouflage, diverse imaging conditions, variability in insect orientation. An end-to-end approach for training deep-learning models, InsectNet, is proposed address these challenges. Our has following key features: (i) uses a large dataset images collected through citizen science along with label-free self-supervised learning train model, (ii) fine-tuning this model using smaller, expert-verified regional datasets create local (iii) which provides high prediction accuracy even species small sample sizes, (iv) designed enhance trustworthiness, (v) democratizes access streamlined machine operations. This global-to-local strategy offers more scalable economically viable solution implementing advanced systems across ecosystems. We report accurate (>96% accuracy) numerous agriculturally ecologically relevant species, pollinators, parasitoids, predators, InsectNet fine-grained identification, works effectively challenging backgrounds, avoids making predictions when uncertain, increasing its utility trustworthiness. The associated workflows are available web-based portal accessible computer or mobile device. envision complement existing approaches, be part growing suite AI technologies addressing

Language: Английский

---

Parameterization before Meta-Analysis: Cross-Modal Embedding Clustering for Forest Ecology Question-Answering

Forests, Journal Year: 2024, Volume and Issue: 15(9), P. 1670 - 1670

Published: Sept. 22, 2024

In the field of forestry ecology, image data capture factual information, while literature is rich with expert knowledge. The corpus within can provide expert-level annotations for images, and visual information images naturally serves as a clustering center textual corpus. However, both represent large rapidly growing, unstructured datasets heterogeneous modalities. To address this challenge, we propose cross-modal embedding clustering, method that parameterizes these using deep learning model relatively few annotated samples. This approach offers means to retrieve relevant knowledge from database through question-answering mechanism. Specifically, align across modalities pair encoders, followed by fusion, feed into an autoregressive generative language user feedback. Experiments demonstrate enhances performance recognition, retrieval, models. Our achieves superior on standardized tasks in public question-answering, notably achieving 21.94% improvement task ScienceQA dataset, thereby validating efficacy our approach. Essentially, targets combining perspectives multiple utilizing representation text. effectively addresses interdisciplinary complexity ecology parameterization encapsulating species diversity conservation images. Building foundation, intelligent methods are employed leverage large-scale data, providing research assistant tool conducting ecological studies larger temporal spatial scales.

Language: Английский

---

Ethical issues in lethal insect monitoring

Current Opinion in Insect Science, Journal Year: 2024, Volume and Issue: 66, P. 101279 - 101279

Published: Oct. 5, 2024

Language: Английский

---