PROJECTS
These are the most representative projects throughout my academic career, each project presents a link available for the purpose of deepening the information about it. Any doubt or interest please do not hesitate to contact me!
Latest Projects
Line - Following Autonomous Robot
Aim
Implement digital PID control with fast response.
Description
I led a group of 5 students in the developing of a control algorithm where taking as input the reflectance of the running circuit (measured by sensor qtr8), the autonomous robot will adjust the speed of both motors to reach the setpoint. This set point is the same set point which is achieved by measuring the reflectance of the sensors located in the middle of the robot.
Result
Line-following autonomous robot that participated in ROP FEST-Bogotá(2019), ROP FEST-Neiva (2019), Universidad Antonio Nariño - Neiva (2019), EXPO CORHUILA - Neiva (2019) where in the latter it reached the first place as the fastest car that completed the race circuit in the shortest possible time.
Aim
Generate a tool for students to facilitate the process of finding the exact resistance for circuit construction.
Description
Trained a YOLO V3 model using own dataset by taking and manually tagging more than 500 pictures of 3 different classes and using Roboflow platform for data augmentation up to 1500 pictures. The CNN was able to detect 3 different categories of resistor (1 KΩ, 2 KΩ and 330 Ω common resistors used in academy).
Result
achieved a model with 70% accuracy in class detection, however, the idea is to create a larger dataset and do a better tuning of the parameters to increase this accuracy. In addition, the plan is for this model to run behind a web page that allows smart devices through the camera to recognize the resistors.
Aim
To develop a code which is capable of detecting faces
Description
The development is focused in generate a bounding box, as well as a face count, on the image as much as possible in real time when a face is detected, based on the Opencv library in conjunction with its cascade classifier.
Result
Tool that allows to detect the position of the face in real time of any person in front of the camera without the need of having a machine with a great processing. Also this project This project gave me a great interest in computer vision and deep learning.