CHALLENGE

Greenedge Contest: Energy Aware Image Classification

In this competition, we are calling on participants to develop a model that excels in both energy efficiency and accuracy when classifying the TinyImageNet test set. Specifically, the participants will tackle a multi-class classification problem on this dataset, focusing on optimizing the energy-accuracy trade-off during the inference phase. We’re eager to see innovative approaches that effectively balance minimizing the average energy consumption with delivering accurate results.

Useful Links

• Competition Description
• Example Notebook
• General Description for Greenedge Contests
• Registration Form

Who Can Participate?

BS, MS level and PhD students are invited to participate in the challenge. The submitted projects will undergo an evaluation by a team of scientists from the GREENEDGE consortium. The winners will be invited to participate in the final GREENEDGE workshop where they will present their work and will receive a prize. The final workshop will be co-located with an international conference and will take place around September 2024 (detailed instructions will follow). You can participate via this link.

Participation Deadline: May 15, 2024

Webinar

We will be holding a webinar on 22nd May 2024. More details about the webinar will be shared soon. We will meet the teams, providing guidance and help in the definition of the objectives, clarifying any possible questions that the teams may have.

Participation Instructions

• Teams of up to 3 people are allowed.
• Subscribe to the challenge using the form at this link.
• Final project submission from the teams: June 15, 2024. (A submission link will be sent to the participants in due time).
• Contest winners’ announcement: July 30, 2024.

Evaluation Phase

An evaluation report will be generated for each received solution. The final score will be decided by a review committee. The evaluation will be based on the following criteria:

• adopted methodology (10 points);
• achieved results (10 points);
• code (8 points);
• presentation (8 points).

Acknowledgements

The present work has received funding from the European Union’s Horizon 2020 Marie Skłodowska Curie Innovative Training Network Greenedge (GA. No. 953775).

Contact Information

To ask questions, please feel free to create an issue via the Issues tab.

We look forward to your participation!