Examinando por Autor "Fernandez-Gallego, Jose A."
Mostrando1 - 4 de 4
Resultados por página
Opciones de clasificación
- PublicaciónAcceso abiertoChained deep learning using generalized cross-entropy for multiple annotators classification(2023-03-16) Triana-Martinez, Jenniffer Carolina; Gil-González, Julian; Fernandez-Gallego, Jose A.; Lugo González, Carlos AndrésSupervised learning requires the accurate labeling of instances, usually provided by an expert. Crowdsourcing platforms offer a practical and cost-effective alternative for large datasets when individual annotation is impractical. In addition, these platforms gather labels from multiple labelers. Still, traditional multiple-annotator methods must account for the varying levels of expertise and the noise introduced by unreliable outputs, resulting in decreased performance. In addition, they assume a homogeneous behavior of the labelers across the input feature space, and independence constraints are imposed on outputs. We propose a Generalized Cross-Entropy-based framework using Chained Deep Learning (GCECDL) to code each annotator’s non-stationary patterns regarding the input space while preserving the inter-dependencies among experts through a chained deep learning approach. Experimental results devoted to multiple-annotator classification tasks on several well-known datasets demonstrate that our GCECDL can achieve robust predictive properties, outperforming state-of-the-art algorithms by combining the power of deep learning with a noise-robust loss function to deal with noisy labels. Moreover, network self-regularization is achieved by estimating each labeler’s reliability within the chained approach. Lastly, visual inspection and relevance analysis experiments are conducted to reveal the non-stationary coding of our method. In a nutshell, GCEDL weights reliable labelers as a function of each input sample and achieves suitable discrimination performance with preserved interpretability regarding each annotator’s trustworthiness estimation.
- PublicaciónSólo datosHeterosis and reciprocal effects for physiological and morphological traits of popcorn plants under different water conditions(Agricultural Water Management, 2022-01-03) Kamphorst, Samuel Henrique; Amaral Júnior, Antônio Teixeira do; Vergara-Diaz, Omar; Gracia-Romero, Adrian; Fernandez-Gallego, Jose A.; Chang-Espino, Melissa Carolina; Buchaillot, Maria Luisa; Rezzouk, Fatima Zahra; Lima, Valter Jário de; Serret, Maria Dolores; Ortega, Jose Luis ArausIn spite of the benefits of heterosis in maize breeding, little is known about the physiological mechanisms of this phenomenon and its genetic control under different water regimes. This study aimed to understand the heterosis effects on plant growth, the photosynthetic and transpiration traits, and the root traits of four inbred popcorn lines and their hybrids, including their reciprocal combinations. Plants were grown in lysimeters, inside a rain shelter, under two water conditions (water stress – WS; well-watered – WW) until anthesis. Plant growth traits included shoot biomass, plant height, and leaf area. Photosynthetic traits comprised leaf pigment and total nitrogen content, chlorophyll fluorescence, gas exchange, water use efficiency and stomatal index and density, along with the stable carbon (δ13C) and nitrogen (δ15N) isotope compositions of the last developed leaf. Root weight density and specific root length were also recorded. Greater heterosis effects were observed for traits related to plant growth and root weight density, and specifically under WS. Traits related to root weight density in deeper soil layers benefited markedly from heterosis, but there were no advantages in terms of stomatal conductance and water status in general. Apparently, only δ13C supported a better water status under WS, and was observed in the hybrids in particular. Non-additive gene effects were predominant in controlling of most of the growth and root traits studied, supporting the conclusion that the heterosis effect is especially favorable under water-limiting conditions. Moreover, the choice of the female parent is essential for traits related to gas exchange when breeding for better resilience to drought.
- PublicaciónSólo datosImplications of Very Deep Super-Resolution (VDSR) on RGB imagery for grain yield assessment in wheat(2020 Virtual Symposium in Plant Omics Sciences, OMICAS 2020, 2021-09-13) Fernandez-Gallego, Jose A.; Kefauver, Shawn C.; Gutiérrez, Nieves A.RGB imagery has been widely used for crop management practices and phenotyping applications in recent years. Although RGB wavelengths (400-700 nm) are not able to capture all essential plant data (such as with full ultraviolet, near and long infrared wavelength coverage), RGB cameras are the most common types of cameras and are among the versatile imaging devices for proximal remote sensing applications. Deep learning strategies have improved a wide range of processes and deep learning concepts can be included in many applications. This work uses the Very Deep Super-Resolution (VDSP) technique to improve low-resolution RGB images in order to study grain yield assessment in wheat using vegetation indexes. The results show no significant differences between indexes calculated from low-resolution images and low-resolution images processed using VDSP with grain yield.
- PublicaciónSólo datosOpen-Source Software for Crop Physiological Assessments Using High Resolution RGB Images(International Geoscience and Remote Sensing Symposium (IGARSS), 2021-02-17) Kefauver, Shawn C.; Romero, Adrian Gracia; Buchaillot, Ma. Luisa; Vergara-Díaz, Omar; Fernandez-Gallego, Jose A.; El-Haddad, Georges; Akl, Alexi; Araus, José LuísThe state-of-the-art on the use of commercially available consumer color digital cameras, which capture Red, Green and Blue light covering the visible spectrum with broad spectral bands but at high spatial resolution and with accurate color calibration has produced some interesting results in recent years, bringing back the benefits of “hyperspatial” imaging for estimating various plant physiological characteristics related to both biotic and abiotic stressors. Here we will review various RGB vegetation indexes that use the spectral concept for the estimation of biomass and canopy chlorophyll, the Normalized Green Red Difference Index (NGRDI) and the Triangular Greenness Index (TGI), as well as others that are in popular use based on this same concept as more traditional style vegetation indices often used with multispectral data. We will also introduce spectral indexes based on alternate color space transforms such as Hue Saturation Intensity (HSI), CIE-Lab and CIE-Luv and their practical calculations. Practical aspects of the calculation of these RGB vegetation indexes are offered using open-source software plugins for FIJI (FIJI is Just ImageJ), including the MaizeScanner, CerealScanner, and their mobile-to-cloud ODK (Open Data Kit) versions Fusion and CerealsFusion.