Registration
00

day

00

hr

00

min

00

sec

Time to team-up deadline

81

participants

Register

Topic provider

E-Elements Technology, Co. Ltd.

E-Elements was founded in 2005 in Taipei, Taiwan. Headquartered in Taipei, Taiwan,  E-Elements has expanded its footprint and established branch offices in ShenZhen and Nanjin China to service customers in the Asia-Pacific region.

In 2006 E-Elements’ design expertise was recognized and was awarded by Xilinx as their Authorized Training Partner (ATP) for Greater China region. Working closely with the Xilinx University Program (XUP) led by Xilinx CTO office, E-Elements provides customized board design, competitive solution and training to universities and corporations adopting XUP development solution.

E-Elements continues to collaborate with global technology leaders to develop innovative designs and solutions. The company vision is to build the best FPGA products and deliver with unparalleled time to market excellence.


ITRI provides a welcoming work environment that stimulates the full potential of ITRI’s members, allowing them to excel and to perform their best in application-oriented researches. We fully cultivate the capability of independent innovation through participating in international cooperation and academic programs. These collaborative activities are pursued so as to aid in the success of technological incubation and entrepreneur start-ups. We strive to strengthen the technological innovation and value creation ability for Taiwan’s industry in order to remain the competitiveness globally.

Introduction

The Automated Optical Inspection, or AOI, is a fast and accurate optical inspection system based on machine vision technology. The advanced inspection method has multiple advantages over traditional inspections executed manually.

AOI technique can be applied to business in multiple areas such as R&D in high tech, manufacturing, defense, healthcare, environmental protection, and electric utility with great potential to increase productivity with reduced costs and time.

ITRI has dedicated in the R&D of flexible displays for decades and aims to improve accuracy of AOI systems to elevate product quality.

We would like to invite data scientists from all different fields to join this competition. With the provided AOI image data to build a model to identify the correct defect type and strengthen the effectiveness of AOI inspection through data science.

FPGAs (Field Programmable Gate Array) are often used to maximize application performance as workloads and algorithms evolve through reconfigurable fabric. In this topic, the participants are challenged to demostrate their ability to optimize machine learning model for FPGA data center accelerator cards (Xilinx® Alveo™ U50LV). Apart from accuracy of the model, time of model inference will also be taken into consideration.

You have to upload the trained model to Aldea. The platform will load your model to FPGA accelerator card, where the model inference will be performed and scored. Aside from being familiar with AI algorithms, the ability to optimize models for AI accelerator devices such as FPGA cards is equally important in the flourishing future of Edge AI. Come accept the challenge and join the competition now!

 

Participation Flow

  1. Register and download training data
  2. Prepare development environment
  3. Train float model
  4. Quantization
  5. Compilation
  6. Package your model in a zip file. Upload it to Aldea for model inference and scoring

For detailed description, please refer to the Tutorial and Sample Code. Discussions and questions on the forum of this topic are welcomed as well. You may also send email to support@e-elements.com.tw for further questions regarding this competition.

You may rest assured that your submitted model will not be used for any purpose other than the evaluation of this contest.


Prize

The following awards will be given to the top three participants who score at least 60 points on the Private Leaderboard.

First place: PYNQ-ZU development board 1 pc (retail price US$895) + TWCC services equivalent to a maximum of NTD 150,000 based on TWCC pricing

Second place: PYNQ-ZU development board 1 pc (retail price US$895) + TWCC services equivalent to a maximum of NTD 100,000 based on TWCC pricing

Third place: PYNQ-ZU development board 1 pc (retail price US$895) + TWCC services equivalent to a maximum of NTD 50,000 based on TWCC pricing

Honorable mention (multiple winners): PYNQ-Z2 development board 1 pc (retail price US$119) + TWCC services equivalent to a maximum of NTD 20,000 based on TWCC pricing

 

Award Provisions and Eligibility

  • Please note that only the top three participants on the Private Leaderboard who submitted the written report before the deadline are eligible for the award.
  • Participants will not be eligible for the award if no report is submitted before the deadline. (The report will not be disclosed or published)
  • TWCC services can only be awarded to the awardees that hold the nationality of ROC.
  • TWCC services in the award will be valid for six months.
  • TWCC Terms of Service(https://iservice.nchc.org.tw/nchc_service/template/member_contract_3_en.html
  • No awards will be given to the participants that are employees (as well as interns and part-time workers) of ITRI or workers from the topic provider. If the award has been given, the prize will be revoked, and the participants will need to return the prize money as well as certificate.

 

Activity time

All time are based on UTC+8.

TimeEvent
2021/08/27Registration starts
2021/08/27Uploading begins
2021/11/17 23:59:59Upload deadline
2021/11/19Private Leaderboard announcement
2021/11/23 23:59:59Report submission deadline
2021/11/30Awards announcement

 

Evaluation Criteria

Apart from accuracy, speed of model inference as FPS (frames per second) is another criterion for scoring.

The total weighted score is 100 points, with the following distribution:

  • Accuracy ($Accuracy_{weighted}$): 60 points
  • FPS ($FPS_{weighted}$): 40 points


$$Accuracy_{weighted}=\begin{cases}\left ( Accuracy-0.9 \right ) \times 10 \times 60, & \text{ if } Accuracy> 0.9 \\ 0, & \text{ if } Accuracy \leq 0.9\end{cases}$$

$$FPS_{weighted}=\begin{cases}\frac{FPS}{1000}\times 40, & \text{ if } FPS < 1000 \\ 40, & \text{ if } FPS \geq 1000\end{cases}$$

Rules

  • The evaluation will be based on the final uploaded result. If two teams (or more) get the same evaluation scores, the time of uploading will determine the ranking.
  • The maximum number of uploading times is 3 times per day.
  • This topic does not offer a team-up option, only one account per person is allowed, and each person can only participate once. If any violations are found, people who are involved would be forced to withdraw the activity.
  • All the data, techniques and source codes that are used belong to the participants. If any third-party claims their intellectual property rights or other rights and interests are being violated, the participant will need to handle the disputes personally. If any participants violate intellectual property rights, they will be disqualified, and shall bear legal responsibilities.
  • All the achievements and their IPRs (intellectual property rights) belong to the participants, and the Copyright License Agreements, patent applications, technology transfer and equity distributions of them, should be in accordance with the relevant Laws and Regulations.
  • The organizers reserve the right to enquire the results or take any related actions.
  • There should be no answer discussions between different accounts, or it will be considered as cheating.
  • If there is cheating or fraud during the activity, the participant that cheats will be disqualified from the activity and the vacancy would be filled up by other participants in the ranking order.
  • After uploading, the answers of test data would be divided into two parts to calculate the score:
    • Before the deadline of the activity: The system will examine and calculate the score refer to the Ground Truth of approximately 40% of the entire test data. The result will be posted on the Public Leaderboard.
    • After the deadline of the activity: The system will examine and calculate the score referring to the Ground Truth of the remaining test data (approximately 60%). The result will be posted on the Private Leaderboard as reference for the final score and ranking.
  • Any artificial marking is forbidden.
  • No awards will be given to the participants that are employees (as well as interns and part-time workers) of ITRI or workers from the topic provider. If the award has been given, the prize will be revoked, and the participants will need to return the prize money as well as certificate.
  • Should there be disputes, the organizers reserves the right to the final decision.
  • The organizers reserve the right to modify any details regarding the contest when needed.