Computer Vision
Synthetic Data Generation: Autonomous vehicles rely on annotations for object detection and image segmentation task. This pipeline tries to solve this problem, where you use as an input a CAD model and the output is a realistic image with its annotation. (In progress: this is an improvement to my work at Evocortex)
Open source github repo: vanttec_synthdata
Image classification for glaucoma (eyes disease): image classfication using firstly a segmentation model and secondly a classification model for glaucoma diagnosis. Project for the Hospital de la Ceguera
Autonomous Vehicle
Unmanned Underwater Vehicle: VantTec’s UUV is a project at the Tecnológico de Monterrey University. I lead the UUV project one year before my engineering graduation and opened up multiple projects in parallel. The results here shown were made entirely during my time.
Open source github repo: vanttec_uuv
Next project
Real-time image enhancement for underwater applications: The UUV can’t rely more on its camera, because the water’s attenuation coefficients. Methods such as Sea-Thru require a couple of seconds and a depth map which makes it unfeasible for a real-time application.
Explainable AI for Synthetic Data Generation: Implementation of XAI methods for visualizing the object detection flaws, with a focus on improving the synthetic data generation strategy. (This is an improvement to my work at Evocortex)