Exploratory studies are presented here that have not been the
subject of any scientific enhancement or advanced development, but with strong potential, both scientific and applicative.These studies focus on « innovative » sensors and new analytical methods.
The sensors
This is an « in-depth » technology watch designed to assess the interest of certain devices for the Group’s research subjects, in particular:
- NirGreenBlue hybrid sensors for « measuring » the chlorophyll activity of trees; read more…
- Plenoptic sensors for introducing depth maps into classification processes; read more…
- The mobile « Lidar » backpack platforms for high-speed digitisation of urban or peri-urban environments; read more…
Contact: Philippe Borianne
Analytical methods
These studies propose to evaluate in detail the principles of the methods and present some preliminary results
2017 : deep neural network for fruit counting
This work is initiated as part of the PixYield action led by Emile Faye, an agroecologist working on the performance of horticultural cropping systems. It allows the group to deploy and specialize the CNN Faster Region-based network to identify and locate mangoes in color images; read more…
Contact: Philippe Borianne
2017 : deep neural network in remote sensing
This work focuses on the identification and localization of coconut palms in aerial color images of natural forest using the Faster Region-based CNN network . Its main interest is to identify the scientific and technical locks specific to the processing of aerial forest cover data by convolutional neural network. It prefigures the orientations of the CocoDeep project formalized in early 2018; read more…
Contacts: Frédéric Borne, Gaëlle Viennois
2014 : low cost infrared sensor for spectral analysis of trees and plants
Scientific animation action (12 to 18 months) around a multidisciplinary workshop financed by the Numev lab (Digital Solutions, Materials and Modelling for the Environment and the Living) and led by the Lirmm. This scientific animation project consists in setting up a multidisciplinary scientific reflection workshop in order to implement a low cost infrared sensor for the spectral analysis of plants and trees. This project brings together specialists in sensor (IES), visual data processing (LIRMM) and plant modelling (CIRAD) to study the characteristics of « low cost » acquisition devices read more
Contact: Philippe Borianne