Pollination is critical to many ecosystem services, such as the maintenance of biodiversity and human nutrition. Addressing how plant-pollinator interactions change across environmental gradients requires quantifying pollinator visitation. However, we are limited by the available tools.
Ecologists typically observe pollinators in the field, record the species identity of the easily distinguished individuals, and collect with a net the remaining individuals so that they can be later identified to species in the lab. This is labor and time intensive, and thus prevents the massive data collection that is necessary to achieve a general, predictive understanding of how pollination responds to global change.
Therefore, we want to use machine learning methods and technology to automatize ecological data collection on plant-pollinator interactions.
A first step is to develop a camera system capable of capturing clear images of visiting insects on flowers. This can be as simple as using a digital camera for time-lapse photography with enough battery supply, capturing images every second. We want to explore with such a tool (Fig .1) in the summer of 2021 with the goal of capturing insect images. The insects from the collected images will be identified by an entomologist down to lowest possible taxonomic level (order > family > genus > species).
We then want to use the time-lapse images for testing an object detection model (e.g.: image contains insect or not and draw the bounding box).
The second part of the project is about using the captured images and images from existing online databases (like iNaturalist and Observation) for training state of the art Convolutional Neural Networks (CNN) with the goal of insect identification down to the lowest possible taxonomic level (Fig. 2).
Finally, the developed CNN models together with a tailored camera system will be used for pollinator identification in field studies aimed at quantifying plant-pollinator networks. We also want to derive abundance data about each pollinator. We will focus on certain families from four orders of pollinators: Coleoptera, Diptera, Hymenoptera and Lepidoptera.
Our goal is not to develop an app, but a tool that can help us obtain insect identity and abundance data about plant-pollinator interactions.
How can you contribute?
If you have images of pollinators and are willing to share them with us, feel free to contact Valentin Ștefan at firstname.lastname@example.org
We also welcome your thoughts, ideas and collaboration.
Steen, R. (2017). Diel activity, frequency and visit duration of pollinators in focal plants: in situ automatic camera monitoring and data processing. Methods in Ecology and Evolution, 8(2), 203-213.
Boussioux, L., Giro-Larraz, T., Guille-Escuret, C., Cherti, M., & Kégl, B. (2019). InsectUp: Crowdsourcing Insect Observations to Assess Demographic Shifts and Improve Classification. arXiv preprint arXiv:1906.11898.
We are looking for BSc/MSc students in Leipzig for a HiWi position max 19 h/week for one year - Capturing and labeling field images of insect pollinators on flowers. Deadline 23 May 2021. See more detais here.