[PhD’s Corner] Kaaviya Velumani: Continuous monitoring of crop development from IoT sensors fixed in the field

Kaaviya is one of the #DigitAg labelled PhDs

Continuous monitoring of crop development from IoT sensors fixed in the field


  • Start Date:  September 2018
  • University:  Avignon University
  • PhD School: A2E – ED 536 Agrosciences & Sciences
  • Field(s): Agronomy, Computer sciences
  • Doctoral Thesis Advisor: Frédéric Baret, Inra EMMAH Avignon et Raul Lopez lozano, Inra EMMAH
  • Co-supervisors : Frédéric Baret, Inra EMMAH Capte
  • Funding: Cifre Hiphen
  • #DigitAg: ,Axis 5:Data Mining, Data Analysis and Knowledge Discovery, Challenge 2: Digital solutions to optimize the genotype in changing production systems and markets

Keywords: IoT, time series, image processing, crop phenology, deep learning, organ detection, spatial representativeness


The recent development of field sensors and high-throughput platforms offers new opportunities to acquire near-real time, and valuable information on plant dynamics. These new systems have an enormous potential as instruments for decision support systems in agriculture.

In this context, the objective of this thesis is to propose a methodological framework that permits to exploit a new prototype of a field system named IoTA –Internet of Things for Agriculture– developed by Bosch Inc. with the contribution of CAPTE in the domains of precision agriculture and high-throughput phenotyping.

To achieve that goal, the thesis will first investigate how the raw data provided by the IoTA devices –a RGB camera, two PAR sensors to measure transmitted light by the canopy, and a multi-spectral radiometer– can be exploited to provide reliable information of highly valuable physiological traits. These traits include the dynamics of vegetation structure (e.g. the green fraction or leaf area index), the detection of key phenological stages (e.g. heading date), the density and size of reproductive organs, and the presence of diseases. Several algorithms such as deterministic models and deep learning approaches will be validated against ground data to identify a set of algorithms that can be suitable for their use in an operational context.

Secondly, the sampling strategy that will permit network of IoTAs to acquire representative observations of crop canopies at different spatial scales will be also investigated. The footprint of the IoTA sensors is few square meters, and their operational use on phenotyping experiments (thousands of microplots of 20 m2), commercial fields (several hectares), or over a territory requires an optimal strategy to distribute the systems. That will permit to develop empirical transfer functions that could support observations with wide spatial coverage given by UAV or a high-resolution satellite (e.g. Sentinel 2).


Contact:  kaaviya.velumani [AT] inra.fr​

Social Networks : LinkedIn – ResearchGate

Communications /Publications

Kaaviya Velumani, SimonMadec, Benoit de Solan, Raul Lopez-Lozano, Jocelyn Gillet, Jeremy Labrosse, Stephane Jezequel, Alexis Comar, Frédéric Baret (2020), An automatic method based on daily in situimages and deeplearning to date wheat heading stage , Field Crops Research – https://www.sciencedirect.com/science/article/abs/pii/S0378429019321604?via%3Dihub