Bokeh Effect Project
Hamesh is a Master’s degree applicant who just completed his Bachelor’s studies at SRM University, Tamil Nadu, India. He is interested in photography and computer science, hence his choice of project focused on the bokeh effect.
Hamesh has just completed his Bachelor’s degree in the field of Computer Science and Engineering. As an avid programmer and a Masters degree aspirant, I am keen on learning and developing my skill sets by exploring the various applications in the field of Machine Learning and AI.
Scope of the program
- Lay out expectations, decide on a project, implementation, testing and analyzing results
- Acquire advanced machine learning skills and conducting research in machine learning
What challenges did he face?
- Hamesh needed a stronger portfolio to apply for his desired master’s program in the United States.
How did our program help him?
- Hamesh gained a deeper understanding of machine learning in terms of bokeh effect and convolutional neural networks at the master’s level.
- His hardwork in this program has earned him the mentor’s trust and recommendations for master’s application. Hamesh’s mentor from Duke University submitted 6 recommendation letters to different universities.
From the perception of an image by the computer to Deep Learning models for Image Restoration, the Mentorship Program had taught me various new skills in the field of Computer Vision and Deep Learning. I had the opportunity to work under the guidance of my mentor, who introduced me to the current state-of-the-art models in Computer Vision.
We worked on a project to render bokeh effect on a monocular image using Vision Transformers. We designed an encoder-decoder architecture with Transformers as the backbone for the encoder to work on patches of the entire image, resulting in more globally coherent predictions. This work establishes state-of-the-art results along with fast prediction by denouncing the use of any Depth Estimations.
Not only was I able to learn new and exciting concepts relevant to the field, but also gained new skills with regard to conducting research. With my mentor’s advise, I learned the basic approach to reading and understanding a research paper and replicating the work from scratch. This provided a clear understanding of the models and room for contributions or improvements, if any.
The unique and fun teaching methods, and the interactive session design of the program has surely helped me gain more experience in the field of my interest and move forward in my career.Hamesh Narzary
Matched professor in Computer Science:
An Assistant Professor in the Department of Electrical and Computer Engineering and a Faculty Network Member of the Duke Institute for Brain Sciences. He holds a number of certificates in education, networking and IT, including Cisco’s CCNP, CCNA, and ITE Instructor Trainer.
“Hamesh is a very motivated and hard-working student. What I like most about him is his passion to learn more and never giving up.“Assistant Professor from Duke University