GSoC 2021 Proposal: YOLO models on the X15/AI

Jump to: navigation, search


YOLO models on the X15/AI[edit]

{{#ev:youtube|Jl3sUq2WwcY||right|BeagleLogic}} About Student: [1]
Mentors: Hunyue Yau
Code: not yet created!
Wiki: http://elinux.org/BeagleBoard/GSoC/ProposalTemplate <- for now in a personal discussion board
GSoC: [ttps://elinux.org/BeagleBoard/GSoC/Ideas-2021#YOLO_models_on_the_X15.2FAI]

Status[edit]

Discussing the tentative ideas with Hunyue Yau on IRC.

Proposal[edit]

Please complete the requirements listed on the ideas page and fill out this template.

About you[edit]

IRC: jduchniewicz
Github: [2]
School: University of Turku/KTH Royal Institute of Technology
Country: Finland/Sweden/Poland
Primary language Polish
Typical work hours (We have mentors in various time zones): 8AM-5PM EET
Previous GSoC participation: Participating in GSoC, especially with BeagleBoard would further develop my software and hardware(BB X15/AI architecture) skills and help me apply my current knowledge for the mutual benefit of the open source community. I aim to deliver a component which will be usable in many upcoming releases of YOLO model and hopefully other models.

About your project[edit]

Project name: YOLO models on the X15/AI (with an extensible interface for other models)

Description[edit]

The main idea of the project is to accelerate Deep Learning models with help of available hardware resources on the BB X15 and BB AI platforms. Current inference times are abysmal for any real-time (or even slightly laggy but bearable) application ranging from 15 to 35 seconds per frame. This is unacceptable and this project will alleviate this problem and enable efficient deployment of other models once the Texas Instruments Deep Learning library allows for that (RNNs, LSTMs and GRUs are planned to be released).

The main scope of the project is to deploy YOLOv3 (tiny?) model and utilize the TIDL library C++ bindings to make it run smooth and efficient. In 10-20 sentences, what are you making, for whom, why and with what technologies (programming languages, etc.)? (We are looking for open source SOFTWARE submissions.)

Timeline[edit]

Provide a development timeline with a milestone each of the 11 weeks and any pre-work. (A realistic timeline is critical to our selection process.)

Mar 29 Applications open, Students register with GSoC, work on proposal with mentors
Apr 13 Proposal complete, Submitted to https://summerofcode.withgoogle.com
May 17 Proposal accepted or rejected
Jun 07 Pre-work complete, Coding officially begins!
Jun 17 Milestone #1, Introductory YouTube video
June 24 Milestone #2
June 30 Milestone #3
July 12 18:00 UTC Milestone #4, Mentors and students can begin submitting Phase 1 evaluations
July 16 18:00 UTC Phase 1 Evaluation deadline
July 23 Milestone #5
July 30 Milestone #6
Aug 06 Milestone #7
August 10 Milestone #8, Completion YouTube video
August 16 - 26 18:00 UTC Final week: Students submit their final work product and their final mentor evaluation
August 23 - 30 18:00 UTC Mentors submit final student evaluations

Experience and approach[edit]

In 5-15 sentences, convince us you will be able to successfully complete your project in the timeline you have described.

Contingency[edit]

What will you do if you get stuck on your project and your mentor isn’t around?

Benefit[edit]

If successfully completed, what will its impact be on the BeagleBoard.org community? Include quotes from BeagleBoard.org community members who can be found on http://beagleboard.org/discuss and http://bbb.io/gsocchat.

Misc[edit]

Please complete the requirements listed on the ideas page. Provide link to pull request.

Suggestions[edit]

Is there anything else we should have asked you?

05:49, 28 March 2021