DAIRNet is pleased to promote the Workshop on Complex-Valued Deep Learning. The workshop will also host SARFish challenge.
The governments of Australia, the United Kingdom, and the United States are pleased to announce a workshop on complex-valued deep learning as part of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2024). The workshop will promote research and discussion on the detection and classification at global scale making use of complex-valued networks, multimodal representation learning methods and natural spacing processing techniques. It will also provide a widely available dataset and benchmark focusing on the joint use of GRD and SLC in synthetic aperture radar imagery (SAR), promoting and showcasing the latest research to combat illegal, unregulated, and unreported fishing. Call for papers is available here.
The workshop will also host the SARFish challenge. SARFish is a free and open large-scale complex-valued SAR dataset for the identification of vessels involved in illegal and unregulated fishing. The dataset can also be used to advance the state of the art in automated detection from SAR imagery, contextual representation learning in GRD-SLC data and deep complex-valued networks.
The SARFish dataset will be used in a challenge to stimulate developments on other topics of interest at the workshop that can naturally tackle complex-valued data such as quantum-inspired approaches.
Topics of Interest
These include but are not limited to:
• Complex-valued deep networks
• Quantum-inspired machine learning methods for complex data
• Scene understanding and semantic labelling in global-scale marine environments using GRD-SLC SAR
• Feature representation, indexing and analysis of SAR images
• Multi-scale methods for detection of small maritime targets in SAR images
• Representation learning for GRD-SLC SAR
• Natural spacing processing for GRD-SLC SAR data
• Open-set recognition and detection for SAR
• Large-scale object detection and recognition.
• Semi-, weakly-, self-supervised learning for complex-valued data
• Few-shot algorithms for SAR and other complex-valued data modalities
• Data-efficient neural architectures for SAR
• Transfer learning and domain adaptation in complex-valued networks
Papers presented at the workshop will be published as part of the WACV Workshops Proceedings, hence refer to the WACV author guidelines. All submissions will be handled electronically via CMT (the submission site is available here).
Papers are limited to eight pages, including figures and tables, in the WACV style. Additional pages containing only cited references are allowed. Papers that are not properly anonymized, or do not use the template, or have more than eight pages (excluding references) will be rejected without review.
For further information, please contact firstname.lastname@example.org
The Workshop on Complex-Valued Deep Learning and SARFish Challenge is proudly supported by the Defence Science and Technology Group (Australia), Defense Innovation Unit (USA), Defence Science and Technology Laboratory (UK), and the Australian Institute for Machine Learning (Australia).