- Contest Sponsored by Xilinx for Kshitij’2019
To build an application using Python in conjunction with the Xilinx PYNQ platform and share back to the PYNQ Community.
Depends on your creativity, e.g. IoT, Video Processing, Any compute heavy workload being accelerated.
The best performers will be awarded with various tokens of appreciation from Xilinx. These items are listed below.
-
Top three best apps will be awarded
- First prize of Rs 45,000/-
- Second prize of Rs 30,000/-
- Third prize of Rs 15,000/-
-
Consolation cash prize for every shortlisted idea which is then showed to be working on hardware too.
-
Icing on the cake is that the institutes of shortlisted teams get to keep the PYNQ-Z2 board!
- Form a team (of Max 4)
- Come up with an idea to solve a problem
- Write up an abstract: a. Demographics: Team name, Branch, College, Team members skills background etc. b. A single page to describe: Problem statement, Solution proposed (Include Python code compiled on CPU at this stage or proposed flowchart of algorithm) c. Supporting Block Diagrams (if needed) – limited to 2 diagrams
- Abstract entries will be evaluated at Xilinx and around 10 teams will be shortlisted. The shortlisted Institute will be receiving PYNQ-Z2 platform board (based on a request from an Institute/Prof/Dept) – in order to take the ideas onto hardware
- Shortlisted teams would be required to bring their idea to life using PYNQ. Submit a 2 minute YouTube (or similar) video of your project. Working source code with results also to be pushed to Github on or before 17th Jan’2019.
- The submitted apps will be evaluated based on the judging criteria and winners will be announced on the finale event of Kshitij’2019
- Creative Problem Identification
- Python (working knowledge)
- Domain Knowledge (of the problem space identified)
- Hardware Knowledge (to be able to understand hardware resources available on PYNQ)
| Level | Task | Deadline |
|---|---|---|
| 1 | Abstract Submission | 15th Dec'18 |
| 2 | Abstract Acceptance Notification | 19th Dec'18 |
| 3 | Final Source Code Submission | 17th Jan'19 |
| 4 | Winners Declaration | On Kshitij'19 Fest |
| Score (1-10) | Criteria | Description | |
|---|---|---|---|
| Idea | 10 | Creativity | How creative or innovative is the idea behind the design? |
| 10 | Viability | How relevant is the design's use case to a real problem? Is the python code working on CPU in the first place? Or does the flowchart lead towards a valid solution? | |
| Open Source and Reproducible | 10 | Open-Source | Are build steps provided, python packaged, available on Github as notebook and source code. |
| 10 | Reproducible | Could someone reproduce this work starting from a pynq v2.3 image and going through their notebook. | |
| Implementation | 10 | Feasibilty | Does it actually work? If you can find a suitable existing overlay form the PYNQ community, we recommend to resuse for faster bring up (http://www.pynq.io/community.html) |
| 10 | Uniqueness | How well does the app leverage a variety of different open-source APIs and programmable logic? |
- Abstract Evaluation will be judged on Idea category only.
- The shortlisted abstracts will be judged on remaining categories.
- PYNQ being open sourced project of Xilinx, much of the information is available online: http://www.pynq.io/
- Checkout the Presentations and Quickstart sections in this repo
- One pre-determined slot one hour per week, given by Xilinx Employee for support/brainstorming
-
We received immense response of 32 abstracts from 11 different engineering institutes across India. Of these, 21 abstracts are from IIT Kharagpur alone. We saw the teams have come up with relevant applications in the domains like Security, Home Automation, Computer vision, Smart Irrigation, Smart Traffic management, Li-ion battery management using Machine Learning(ML), Automotive Advanced Driver Assistance Systems(ADAS) and other ML applications.
-
With many high quality submissions selecting 10 abstracts was indeed a challenge. Hence the judging panel comprising of technical experts has come up with 16 shortlists. Of which 11 shortlists from IIT KGP are requested to share the 6 boards being provided. Institutes of shortlisted teams will get to keep the PYNQ-Z2 board.
-
The shortlisted teams are requested to consult their professor in Electronics/Electrical/Computer Science/Mathematics Department and provide the official mailing address to radhika.pokanati@xilinx.com on or before 21st Dec'18. The board will be sent only upon request from the professor of the institute. For IIT KGP, all 6 boards will be sent to one professor for easy coordination.
-
Its time to make your awesome idea to work! While you can put efforts to design and build your own overlay(vivado bitstream), it is recommended to use existing overlays if suitable from the PYNQ community for faster bring up (http://www.pynq.io/community.html).
-
You will need to submit a 2 minute YouTube (or similar) video of your application. Working source code with results also to be pushed to Github on or before 17th Jan’2019.
-
The S No below does not indicate the ranking order.
| S No | Team Name | Abstract Idea | Institute |
|---|---|---|---|
| 1 | Proximity Face Detection Lock | Haldia Institute of Technology | |
| 2 | TM190085 | Forest Fires | IIT KGP |
| 3 | Tanishq | Intrusion Detection using Fuzzy Neural Networks | IIT KGP |
| 4 | Trigger word detector | Keyword spotting for offline home automation and speech recognition for unlocking the door | IIT KGP |
| 5 | Virtuality | Military security using Drone | Heritage Institute of Technology |
| 6 | Schwifty Horsemen | Border surveillance | IIT KGP |
| 7 | ASTUTE_FARM | Smart Farming | IIT KGP |
| 8 | Smart Power Monitoring system | Koneru Lakshmaiah Education Foundation (KLEF) | |
| 9 | Muete | Glove with extension | IIT KGP |
| 10 | Rooloo | TRAFYQ - Traffic & Accident Management in India | Sahyadri College of Engineering |
| 11 | reverse_biased | Hand gesture recognition for home automation | IIT KGP |
| 12 | HazeGo | Hardware Accelerated Single Image Super resolution using Xilinx PYNQ Platform | IIT KGP |
| 13 | TM19005D | Advanced ML Based Implementation of Li-ion battery cell balancing and SOC estimation for battery management systems in Electric Vehicles (EVs) using PYNQ platform | IIT KGP |
| 14 | Virgin | Moving object detection using deep learning in conjunction with Xilinx PYNQ platform | NIT Patna |
| 15 | TM19006C | Real-time virtual assistance system for safe navigation without vision | IIT KGP |
| 16 | ROS_Nodes | Eye tracker and braking assistance to avoid crash | IIT KGP |
P.S - We recommend everyone to have unique team name.