Dr. Gilberto Ochoa-Ruiz

email: gilberto(dot)ochoa(at)tec(dot)mx

Official website: http://research.tec.mx/vivo-tec/display/PID_334436

Tecnologico de Monterrey, Computer Science Department - Machine Learning Research Group

General Chair Latinx in Computer Vision at CVPR and ICCV

Bio

Gilberto Ochoa Ruiz is a researcher in Computer Vision, Machine Learning and Internet of Things. He has participated as associated researcher and lecturer in several programs accredited by the CONACYT PNPC program, geared around Computer Science and Communication and Information Technologies. He obtained a maser in Computer Vision in Robotics from Heriot-Watt University and a PhD in Computer Vision and Electronic Imaging from the Universite de Bourgogne (Laboratoire d’Imagerie et Vision Artificielle) He is member of the Sistema Nacional de Investigadores (SNI, Rank I) and of the CONACYT Network on Applied Computational Network (RedICA), the Mexican Societies of IA (SMIA) and Computer Sc. (AMEXCOMP), as well as the Latinx in AI (LXAI) Coalition. He as served as reviewer for ICLR, ICML, CVPR and IJCNN and as general chair of the Latinx in Computer Vision Workshops at LXC @ CVPR and LXCV @ ICCV and as part of the organization or technical program committees of these efforts, as well as other conferences. He is part of the academic staff o the Artificial Intelligence Hub at Tecnologico de Monterrey and he became recently academic partner of the Erasmus Mundus Master in Computational Colour and Spectral Imaging.

His research interests are focused on investigating and implementing novel algorithms and methods for applications in computer vision and medical image analysis. He has ample experience in the design and implementation of smart cameras on FPGA-based Systems-on-Chip for industrial applications (i.e. 3D scanners) at Prefixa Vision Systems and more recently he has become interested in developing optimized machine learning models for edge computing applications.

Research Areas

• Artificial Intelligence (Computer vision and machine learning) for medical imaging

• Applied Computational Intelligence, Smart Connected Devices, AI at the edge

• Optimization techniques for designing edge AI models and devices.

Positions available/Grants

There are available positions (bachelor, master and PhD) for all these projects via CONACYT (Mexican Council for Science and Technology) and other grants for suitable candidates (contact me for further details). International mobility is strongly encouraged and double diplomas are possible. I strongly advice any prospect students to show qualifications in machine deep learning if possible. Showing evidence of previous projects (thesis, articles, GitHub) or qualifications from Coursera (Deep Learning and/or Artificial Intelligence for Medicine) is highly desired.

Collaborators

Gerardo Rodriguez Hernandez, CIATEQ, Mexico

Miguel Gonazalez Mendoza, Tecnologico de Monterrey, Mexico

Andres Mendez Vazquez, CINVESTAV, Mexico

Christian Mata Miquel, Universitat Politecnica de Catalunya

Christian Daul, Centre de Recherche en Automatique de Nancy, France

Sharib Ali, University of Oxford, United Kingdon

Ashutosh Natraj, Vidrona LTD, United Kingdom

Lucile Rossi, Universita di Corsica, France

Lina Maria Aguilar Lobo, Universidad Autonoma de Guadalajara, Mexico

Research Projects

He is currently involved in the following projects related with medical Imaging and digital pathology, win which we are interested in developing new image medical analysis and processing methods, but also machine learning models to facilitate the diagnosis of various diseases. We are also actively exploring the development of interpretability and explainability tools for deep learning models.

Robust Surgical Tool Segmentation, Tracking and Depth Perception

In collaboration with Dr. Sharib Ali from the Department of Engineering Science, Institute of Biomedical Engineering of the University of Oxford (United Kingdom)

Goal: To develop new datasets, schemes and models for implement robust and real-time computer vision methods for Computer Integrated Surgery (CIS) applications and procedural quality assessment purposes

RECONDITE: Deep learning and image analysis methods for improving the endoscopic identification of kidney stones composition

In collaboration with Prof. Christian Daul the Centre de Recherche en Automatique de Nancy, CRAN (France) and the Institut National de la Santé et de la Recherche Medicale (INSERM)

Goal: To investigate deep learning algorithms for automatically classifying in vivo kidney stones from endoscopy images

PROTEUS: Endoscopic 3D View Enhancement and Automatic Categorization of Gastro-Intestinal Inflammations from Endoscopic data

In collaboration with Prof. Christian Daul the Centre de Recherche en Automatique de Nancy (CRAN) and the Hopital Ambroise-Paré (Paris)

Goal: Real-time 3D organ reconstruction from epithelial surfaces, such as those found in the esophagus, stomach and colon. To explore DL-based SLAM for obtaining fine-grained organ atlas for cancer diagnosis and to combine it with methods for automatically categorizing lesions or pre-cancerous areas from endoscopic procedures

ISOLATE: SegmentatIon and claSsification Of vascuLar pATtern symmEtries on cerebral vessels using DL

In collaboration with Dr. Christian Mata and Prof. Enrique Benitez from the Biomedical Engineering Research Center (CREB, Barcelona) of the Universitat Politecnica de Catalunya (Spain) and the Hospital Sant Joan de Deu (Barcelona)

Goal: To develop novel CADx tools for aiding physicians in the diagnosis of CP. Various algorithms for vessel segmentation and skeletonization have been explored and tested. The results of these preprocessing methods are to be used for classifying vascular pattern asymmetries

Watch: Wildfire Analysis Through Computer vision tecHniques

A collaborative project for early wildfire dentification and fire widespread forecasting. A collaboration with several Mexican universities and the Universitá di Corsica (France) with Prof. Lucile Rossi from Laboratoire Sciences Pour l’Environnement and the GOLIAT team

Goal: To investigate novel drone-IoT model for detecting, monitoring and spread forecasting of fires using computer vision

MEANING MEtric And maNIfold learNinG

A collaboration with CINVESTAV Guadalajara with Prof. Andres Mendez-Vazquez group.

Goal: To investigate novel methods for deep metric learning, and related areas such as few-shot learning (FSL), OOD generalization, neural network dissection and explainability.

Students

PhD

Mansor Ali Teevno, Robust Surgical Tool Segmentation, Tracking and Depth Perception

Daniel Flores Ariza, Integrating causality in the interpretability of artificial intelligence models applied to medicine Github Rep

Jorge Zapata, GEMINI: Guided Metric Learning [GitHub Rep]

Ricardo Abel Espinosa Loera, Endoscopic View Enhancement using Deep Learning-based 3D Reconstruction Techniques

Ivan Reyes Amezcua, Github Rep

Master

Jorge Francisco Ciprian Sanchez, Deep Learning model for early wildfire detection through the fusion of visible and infrared information GitHub Rep

Carmina Perez Guerrero, Characterization of Jet Fire Flame Temperature Zones Using a DeepLearning-based Segmentation Approach GitHub Rep

Daniela Herrera Montes de Oca, Automatic segmentation and classification of vascular pattern symmetries on cerebral vessels using DL

Pedro Esteban Chavarrias Solano, Automatic Categorization of Gastro-Intestinal Inflammations using Deep Learning

Carlos Axel Garcia Vega, Deep Learning Lightinhg Enhancement for Endoscopic Computer Integrated Surgery

Juan Carlos Angeles Ceron, Attention YOLACT++: Achieving robust and real-time medicalinstrument segmentation in endoscopic procedures

Mauricio Mendez Ruiz, Model Extensions for Semantic Segmentation using Few-Shot Learning Approach

Oscar Hinojosa, Automated classification method for ureteroscopic kidney stone images using machine learning GitHub Rep