Embedded Files
La Universidad del Rosario cuenta con amplia trayectoria en programas de pregrado y posgrado en Bogotá. Conoce más ingresando aquí.urosario.edu.co/periodico-nova-et-vetera/nuestra-u/el-ingeniero-oscar-julian-perdomo-charry-nuevo-director-del-programa-de-ingenieria?fbclid=IwAR3kw2kFHC6wZvgCl-vpqdnNwi2UNa4pcTSpkd-uhH3qAsOeGRGjPFPlEzo'Robins' teaches Spanish to deaf children
'Robins' teaches Spanish to deaf children
An interdisciplinary group of Engineering and Health Sciences from the Universidad del Rosario created the first interactive device to support minors with hearing disabilities through literacy. The social robot known as Robins (interactive robot for deaf children), a technology capable of supporting the learning process of reading with an appropriate strategy for children with hearing disabilities. Automatic facial expressions and sign language recognition were programmed into the device to ensure proper interactions. Robins was the only Colombian finalist, in the University Team modality, in the OpenCV AI Competition 2021, sponsored by Microsoft Azure, Intel and OpenCV. This is the world's largest artificial intelligence contest, where researchers managed to position the capabilities of Colombia and Rosario in this area of knowledge.
An interdisciplinary group of Engineering and Health Sciences from the Universidad del Rosario created the first interactive device to support minors with hearing disabilities through literacy. The social robot known as Robins (interactive robot for deaf children), a technology capable of supporting the learning process of reading with an appropriate strategy for children with hearing disabilities. Automatic facial expressions and sign language recognition were programmed into the device to ensure proper interactions. Robins was the only Colombian finalist, in the University Team modality, in the OpenCV AI Competition 2021, sponsored by Microsoft Azure, Intel and OpenCV. This is the world's largest artificial intelligence contest, where researchers managed to position the capabilities of Colombia and Rosario in this area of knowledge.
Patient movement analysis based on artificial intelligence models for a clinical setting
Patient movement analysis based on artificial intelligence models for a clinical setting
The main goal of our proposal centers on a system that integrates hardware and software plus techniques based on artificial intelligence for data analysis allowing intelligent real-time monitoring of physiological variables, activities of displacement, falls, and other physical activities of people with disabilities. Thus, the data analysis should support decision-making in clinical or residential settings. However, the integration and implementation of these algorithms in low-cost hardware, the use of Tiny-ML for real-time running, and the tackling of several unwanted issues in clinical settings still present some open challenges.
The main goal of our proposal centers on a system that integrates hardware and software plus techniques based on artificial intelligence for data analysis allowing intelligent real-time monitoring of physiological variables, activities of displacement, falls, and other physical activities of people with disabilities. Thus, the data analysis should support decision-making in clinical or residential settings. However, the integration and implementation of these algorithms in low-cost hardware, the use of Tiny-ML for real-time running, and the tackling of several unwanted issues in clinical settings still present some open challenges.
Weed Seeker Bot
Weed Seeker Bot
The proposal presents a strategy for the automatic detection of the spatial distribution of weeds in a crop field using deep learning (DL) algorithms applied to RGB-D & multi-spectral images. An image database collected by an unmanned aerial vehicle (UAV) was used to test the performance of our proposed method.
The proposal presents a strategy for the automatic detection of the spatial distribution of weeds in a crop field using deep learning (DL) algorithms applied to RGB-D & multi-spectral images. An image database collected by an unmanned aerial vehicle (UAV) was used to test the performance of our proposed method.
Neural Networks for Detection of Diabetic Retinopathy and Diabetic Macular Edema
Neural Networks for Detection of Diabetic Retinopathy and Diabetic Macular Edema
Given the large number of affected persons, the detection process is expensive and time consuming, thus arising global telemedicine programs for diabetic retinopathy detection. However, DR exists at a very large scale that such programs are not enough for detecting efficiently on a large population basis (12). Moreover, for developing countries these programs are still out of reach, resulting a low screening rate caused by poor access to eye screening services (10). Consequently, millions worldwide continue to experience vision impairment without timely detection, nor proper diagnosis and absence of eye care (12). In order to reduce this sight burden for patients, the number of physicians involved in detection and the cost, the development of a retinal pathology-screening program based to complement telemedicine programs are needed. Various image analysis algorithms based on Neural Networks (NN) have been developed over the last few decades(13). We modified a Neural Networks for DR detection in color fundus images using deep learning methods, addressing the limitations in previously published DR detection algorithms, using a large-scale, heterogeneous real-world fundus data sets, and advantages of being a NN with an automatic characterization of findings on eye fundus images.
Given the large number of affected persons, the detection process is expensive and time consuming, thus arising global telemedicine programs for diabetic retinopathy detection. However, DR exists at a very large scale that such programs are not enough for detecting efficiently on a large population basis (12). Moreover, for developing countries these programs are still out of reach, resulting a low screening rate caused by poor access to eye screening services (10). Consequently, millions worldwide continue to experience vision impairment without timely detection, nor proper diagnosis and absence of eye care (12). In order to reduce this sight burden for patients, the number of physicians involved in detection and the cost, the development of a retinal pathology-screening program based to complement telemedicine programs are needed. Various image analysis algorithms based on Neural Networks (NN) have been developed over the last few decades(13). We modified a Neural Networks for DR detection in color fundus images using deep learning methods, addressing the limitations in previously published DR detection algorithms, using a large-scale, heterogeneous real-world fundus data sets, and advantages of being a NN with an automatic characterization of findings on eye fundus images.
Contact ➡️ e-mail: ojperdomoc@unal.edu.co
Page updated
Google Sites
Report abuse