Abhishek shankar's proposal for Github Externship
Created by: abhishek-a05
Github Externship Proposal
-
Student ID: abhishek_a05
-
Application No -- 31-05_abh665_tfa_291
-
Mentors: Abhishek_ , zeekhuge , jkridner , pratimu
-
Repository: https://github.com/abhishek-a05/Latest-Images
Status
This is Abhishek Shankar's proposal for Github Externship.
About you
- IRC: abhi_shek
- School: Manipal Insitute Of Technology
- Country: Mumbai, Maharashtra,India
- Primary language (We have mentors who speak multiple languages): Hindi, English
- Typical work hours (We have mentors in various time zones): 6-8am and 12-2pm or 6-8pm and 9-11pm IST
- Previous externship and open source experience: I don't have any experience with open source projects
About your project
Project name: Real Time Lane Detection
Description
The purpose of this project is to find a feasible solution to detect road lanes on a low voltage computer that can be easily powered in a regular auto vehicle. The board can be attached to the vehicles dashboard , and a camera facing the road can capture real time video. The Technologies and languages used are Linux Kernel, Python and C.
Timeline
Provide a development timeline with a milestone week and any pre-work. (A realistic timeline is critical to our selection process.)
Date | Milestone |
---|---|
week 1 | Undertanding the RISC-V architecture & studying the board |
week 2 & 3 | Undertanding the linux kernel implementation of the project |
week 4 to 5 | Talking to the mentors |
week 6 to 11 | Developing the model |
week 12 & 13 | Finalising |
Detailed milestone for week 6 to 11:-
- Gathering , cleaning , processing video data
- Choosing a suitable architecture for neural network
- quantizing the model to fit it in the beagle -V starlight
- testing with live camera feed, comparing accuracy , further improvements
Experience and approach
I'm currently pursuing Electronics & Instrumentation in my Pre-final year & pursuing a minor in Embedded systems.I like to work on embedded systems which can incorporate core machine learning/Deep learning techniques. I've previously worked on developing deep learning models for object tracking & detection . I've also worked on Arduino and have developed a mini alarm system , incorporating bluetooth for alerts.
I have a good grip on python , C/C++ , Data structure & algorithms, Linux and deep learning techniques. I believe i will fair well and i will develop and deploy my solution well within time.
What will you do if you get stuck on your project and your mentor isn’t around?
I will read relevant research papers and material to figure out a solution, search youtube or google. I will also consult my friends who have worked on similar projects and try to solve the problem .
Benefit
Real-time lane detection has wide-spread use in intelligent ground vehicles/robots , that are used in remote locations or places where humans cant access such as a nuclear reactor , or a live fire in a building. By developing this model using the beagle-V we can use it as an off the shelf model for lane detection without having to train the model from scratch, and that too with high accuracy.
Conclusion
I would really like to make this project, with the help of your organisation , and actually make a contribution to open-source and mankind. This project can become more than what i envision with the right direction and guidance.