Hello, I’m
Oscar Juarez
Graduated in Image Processing and Computer Vision Master Degree
View my Work
Graduated in Image Processing and Computer Vision Master Degree
View my Work
I am a Mexican national who is passionate about everything related to Tech. In my free time I enjoy gaming and am an avid media and pop culture consumer. In addition to participating in Hackathons on the weekends. Always looking to discover and learn new things, I often find myself enjoying new experiences, especially if they come in the shape of food.
A Mechatronic Engineer by training, I have developed a diverse knowledge base in digital technologies. This includes vocational Computer Vision Engineering experience using Python to develop Deep Learning based technology. I am interested in continuing to expand my knowledge in Deep Learning and collaborate in this cutting edge field. Using creative problem solving to emerge with novel implementations.
13 October 1993
oscar5-juarez@hotmail.com
635775792
Barcelona, Spain
Engineering and problem solving are things I have always been passionate about and have pursued either as student, in hackathons or in my professional experience. This passion has taken me on a path that has involved a lot of learning, including skills in Computer Vision and Deep Learning, and many interesting and high impact projects. Here you can find some of them.
Python programming focused in DataScience but with a complementary skills to perform Tasks like Web Scraping and Pipelines automation.
Strong background in both theoretical and practical knowledge in Deep Learning Models (DNN, CNN mainly), and Machine Learning Models (SVM, Logistic Regression, PCA, etc.)
Enthusiastic person in algorithms and technologies applied to Computer Vision problems. Since the development of my Bachelor Thesis I have not stop to learn about this amazing field.
Because code needs Hardware to be processed. I have a bast experience programming such devices like Arduino, Raspberry Pi and Jetson´s family architectures.
To be able to make the dream come true is needed to validate in a faster way your first design. That is why I am able to design hardware in 3D modeling software such as SolidWorks and materialize it using Laser-cut Machines or 3D printers.
Is important trying to be updated. I am always reading about new technologies and looking to take them into an practical point of view.
Either as student, as Hackathon participant or in my Professional life I am been involved into a very wide variety and exciting projects. From Agriculture, Industrial process or even Art here you can find some of them:
This project aims to recognize and count pests, with the intention of identifying their behavior in order to generate current information and predictions that can help in decision-making. outdoor This was done with a standalone outdor device, it is made up of a 10 MP camera, a solar energy module and communicates through a cellular network This hardware is configured to take daily photographs These images are also integrated into an application that facilitates the classification of the insects found in the traps by experts in the area and that at the same time works to display a summary of the data by date. Using image processing techniques, the area corresponding to the adhesive trap is identified. The identified image is processed by an Artificial Neural Network that is responsible for determining the type of insect.
This project focused on the counting and recognition of assets, in this case a plastic tray where the products are transported. For this challenge, a method called Fast Region-based Convolutional Network (Fast R-CNN) was applied, this method takes as input a complete image and a set of object proposals. Another challenge of this project was the use of wide-angle lenses, due to the limited space that exists between the camera and the entrance to the pit, wide-angle lenses had to be adapted to cover the entire panorama relevant to the recognition. The use of these lenses generates a radial distortion in the image. To correct this distortion, a calibration was carried out on the equipment, finding the intrinsic parameters. The results after this calibration were as observed below:
For Marketing purposes a company asked for the challenge of monitoring the Facial expressions of their customers.We made use of Transfer Learning and fine Tuning to take advantage of a model Trained in IMDB database to perform correctly with the ethnicity of the Mexican Population.
An average user shares a part of her/his life through social media, using open data obtained from Twitter we were able to extract some important features from the personal profile and take that information to offer a hyper-personalized experience at shopping. For that we made use of Computer Vision techniques and Deep Learning to extract the color base of the profile image and use it in some feature over the Web Design of the brand. With the objective of been inclusive we recommended a website with Models from different ethnicities wearing all the pieces of clothes and also consider that characteristic as part of the Hyper-personalization experience.
As part of a campaign to improve the quality of the visitant into the Mexico City Museums we started a campaign called "Be part of the History" where you could copy the Artistic Transfer style of a Famous painting and transfer it to your Photos. Other functionality was the use of a Filter with the face of your favorite Historical Figure. We made use of Deep Neural Networks and Computer Vision techniques to develop the system, the performance can be seen in the images bellow:
This section is always changing, now I am doing my master degree, keep updated and lets talk.
Barcelona, Spain
635775792
oscar5-juarez@hotmail.com
Feel free to contact me, I am sure we will be able to create something amazing together.