Github repo for the ICMI 2025 Grand Challenge
- Organic Waste Sorting
- Robotics
- Multimodal imaging
- AI for Good
The scope of the grand challenge is to explore the recognition of materials through 15-band near infrared (NIR) and single-band thermal images. Multimodal imaging encourages the use of complimentary information to be exploited for a more coherent presentation of scenes. For food waste in particular, information relevant to material is useful.
A double-blind review process will be used to ensure fairness. Each paper will be reviewed by three reviewers. The tentative reviewers are listed below. Model performance will be measured against classification accuracy and segmentation scores such as mIoU and mAP. The validation dataset will be released at a later date than the training data. Code availability may be requested from participants for replicability.
- Matthew Vestal (Matthew.Vestal@canberra.edu.au)
- James Ireland
- Damith Herath
- Xing Wang
- Ram Subramanian
- Roland Goecke
- Yue Ma
- Challenge Site up on April 16th, 2025
- Labelled training set available May 1st, 2025
- Labelled test set available June 1st, 2025
- Paper Submission Deadline: June 15th, 2025
- Notification: July 15th, 2025
- Camera-ready papers: August 15th, 2025
The data can be accessed here: https://drive.google.com/drive/folders/1B4Ax4hbfdX2YpnsU5wNEOr1VqBuoTn2G?usp=drive_link
The validation data can be accessed here: https://drive.google.com/drive/folders/1S-m2zTFl2joPNAXNqHbvQUy2ymvxd9wn
This research is funded by Goterra and the University of Canberra, with in-kind support from Tactical Research Pte Ltd.
A keynote talk will be delivered by a senior employee of Goterra speaking about organic waste management and is associated challenges. Accepted papers will be given the opportunity to present as well.
The accepted papers will be published in the proceedings of conference and the paper format and deadline for submission will follow the same rules reported for ICMI 2025.