Patrice Delmas

Dr. Patrice Delmas


Research Focus & Metrics

Academic Career Length

22Years (2005-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
17 August 2023 in Journal of Food Measurement and Characterization
Verified
Journal Article
Open access
PDF file
MRMitchell Rogers
Mitchell Rogers
JBJacques Blanc-Talon
Jacques Blanc-Talon
MUMartin Urschler
Martin Urschler
PDPatrice Delmas
Patrice Delmas
Over the past two decades, hyperspectral imaging has become popular for non-destructive assessment of food quality, safety, and crop monitoring. Imaging delivers spatial information to complement the spectral information provided by spectroscopy. The key challenge with hyperspectral image data is the high dimensionality. Each image captures hundreds of wavelength bands. Reducing the number of wavelengths to an optimal subset is essential for speed and robustness due to the high multicollinearity between bands. However, there is yet to be a consensus on the best methods to find optimal subsets of wavelengths to predict attributes of samples. A systematic review procedure was developed and applied to review published research on hyperspectral imaging and wavelength selection. The review population included studies from all disciplines retrieved from the Scopus database that provided empirical results from hyperspectral images and applied wavelength selection. We found that 799 studies satisfied the defined inclusion criteria and investigated trends in their study design, wavelength selection, and machine learning techniques. For further analysis, we considered a subset of 71 studies published in English that incorporated spatial/texture features to understand how previous works combined spatial features with wavelength selection. This review ranks the wavelength selection techniques from each study to generate a table of the comparative performance of each selection method. Based on these findings, we suggest that future studies include spatial feature extraction methods to improve the predictive performance and compare them to a broader range of wavelength selection techniques, especially when proposing novel methods.

26 August 2019 in Soil Research
Verified
Journal Article
Closed access
PDF file
CDC. Duwig
C. Duwig
BPB. Prado
B. Prado
ATAnne-Julie Tinet
Anne-Julie Tinet
PDPatrice Delmas
Patrice Delmas
NFNicola Dal Ferro
Nicola Dal Ferro
FMF. Morari
F. Morari
Volcanic soils are important resources because of their unique mineralogical and physical characteristics, and allophanic Andosols represent some of the world’s most fertile soils. However, their unique properties can be lost when cultivated. Most soils in the Central Valley, Mexico, are derived from volcanic materials. This valley encompasses one of the largest water supply systems in the world by volume, but is affected by soil degradation and deforestation. Sustainably managing volcanic soils requires understanding how land use affects their hydrodynamic properties. Gas adsorption and mercury intrusion porosimetry, water retention curves, tension infiltrometry and X-ray tomography were used to describe pore structure characteristics. Two volcanic soils (one Andosol and one derived from indurated tuff – Tepetates), three land uses (maize monoculture, maize–wheat rotation and fallow) and two horizons (Ap and A2 for maize monoculture and maize–wheat rotation) were studied. Tillage affected topsoil by increasing the sand fraction by 38% and decreasing total porosity and macroporosity by 23% and 40% respectively. Macropore size was reduced and the number of isolated macropores was higher in the tilled layer under maize, compared with untilled subsoil. The plot under maize–wheat rotation had lower allophane content, and saturated hydraulic conductivity was reduced by nearly an order of magnitude and water retention by half, compared with maize and fallow plots. Compared with Andosols, Tepetates showed differences in mineralogical composition with lower contents of amorphous compounds and in its porous network characteristics with twice the total and percolating macroporosity compared with the maize plot. Its high content of organic carbon (3.5%) seemed beneficial for its hydrodynamic properties. Sustainable agricultural management of these volcanic soils requires reducing mechanised tillage, avoiding periods when soil is bare, not applying maize–wheat rotation and applying maize–fallow rotation allowing natural vegetation growth.

15 June 2014 in The Photogrammetric Record
Verified
Journal Article
Closed access
PDF file
SBStéphane Bertin
Stéphane Bertin
HFHeide Friedrich
Heide Friedrich
PDPatrice Delmas
Patrice Delmas
ECE. Chan
E. Chan
GGG. Gimel'Farb
G. Gimel'Farb

9 February 2022 in PeerJ Computer Science
Verified
Journal Article
Open access
PDF file
NLNi Liu
Ni Liu
MRMitchell Rogers
Mitchell Rogers
HCHua Cui
Hua Cui
Weiyu LiuWeiyu Liu
Weiyu Liu
XLXizhi Li
Xizhi Li
PDPatrice Delmas
Patrice Delmas
Regular textures are frequently found in man-made environments and some biological and physical images. There are a wide range of applications for recognizing and locating regular textures. In this work, we used deep convolutional neural networks (CNNs) as a general method for modelling and classifying regular and irregular textures. We created a new regular texture database and investigated two sets of deep CNNs-based methods for regular and irregular texture classification. First, the classic CNN models (e.g. inception, residual network,etc.) were used in a standard way. These two-class CNN classifiers were trained by fine-tuning networks using our new regular texture database. Next, we transformed the trained filter features of the last convolutional layer into a vector representation using Fisher Vector pooling (FV). Such representations can be efficiently used for a wide range of machine learning tasks such as classification or clustering, thus more transferable from one domain to another. Our experiments show that the standard CNNs attained sufficient accuracy for regular texture recognition tasks. The Fisher representations combined with support vector machine (SVM) also showed high performance for regular and irregular texture classification. We also find CNNs performs sub-optimally for long-range patterns, despite the fact that their fully-connected layers pool local features into a global image representation.

13 June 2019 in Soil Research
Verified
Journal Article
Closed access
PDF file
IPIlaria Piccoli
Ilaria Piccoli
NFNicola Dal Ferro
Nicola Dal Ferro
PDPatrice Delmas
Patrice Delmas
Andrea SquartiniAndrea Squartini
Andrea Squartini
FMFrancesco Morari
Francesco Morari
Soil organic matter (SOM) plays a key role in soil structure formation, although the bidirectional relationship between SOM and the soil pore network is complex and needs further investigation. Despite great advances provided by X-ray computed microtomography (µCT), it has only been used in a few studies to investigate the organic matter 3D-arrangement within the soil matrix. Results are based on the X-ray linear attenuation coefficient (a), and mixtures of organic and mineral soil fractions could imply overlapping of information that makes any segmentation procedure difficult. The aim of this study was to visualise, segment, and quantify the particulate organic matter fraction (POM) within the soil matrix through X-ray µCT in combination with contrast agents (phosphomolybdic acid and silver nitrate). Two series of repacked soil cores, ‘dry’ and ‘wet’, were scanned through X-ray µCT at a 7-µm resolution. Different segmentation approaches were tested to separate POM from other soil phases: manual, global, and local thresholding methods. Reported algorithms were also compared with a supervised grey value-based (GV) approach followed by morphological operations. Results showed contrast agents increased a of POM, simplifying its identification and the following segmentation on dry cores. The POM was discriminated from the mineral fraction and its content correctly estimated. This was particularly accurate when applying manual thresholding or GV approach with respect to indicator kriging, suggesting that operator-based ability to set threshold level is still the best solution for accurate POM segmentation. Beyond single-phase accounting, different thresholding algorithms and morphological operations also affected POM morphological characteristics. In particular, the simpler was an object shape, the easier was its segmentation. Improvements are thus required to increase the efficiency of automated thresholding algorithms. Moreover, wet cores were exposed to washing-out phenomena that compromised any digital image processing and further POM quantification, implying that more effort should be made to find other suitable staining agents.

24 May 2024 in Future Generation Computer Systems
Verified
Journal Article
Open access
SHShahrokh Heidari
Shahrokh Heidari
MDMichael J. Dinneen
Michael J. Dinneen
PDPatrice Delmas
Patrice Delmas