52nd IEEE Applied Imagery Pattern Recognition Workshop (AIPR 2023)
Sept 27-29th, 2023
Saint Louis, Missouri, USA
Deep Learning Using Synthetic, Augmented, and Natural Datasets

Workshop Chairs: Andrew Kalukin, Kannappan Palaniappan, Michelle Quirk
Program Chairs: Derek Anderson (University of Missouri) and Vasit Sagan (Saint Louis University)

AIPR continues the half-century of success and tradition in pioneering new topics in applied image and visual understanding. Forbes estimates that artificial intelligence (AI) will become a $150 trillion dollar industry. AI is impacting nearly every facet of life and with the advent of large language models and transformers for dialog (i.e. Generative Pre-trained Transformer (GPT), Bard) and novel image or video generation (i.e. Dall-E) it will likely redefine our world faster than previous advances in computational learning. While machine learning (ML) and deep learning (DL) is heavily anchored in supervised learning, recent algorithms (e.g., ChatGPT, DALL-E, etc.) are using self-supervised, transfer, and reinforcement learning. However, all these approaches are data intensive. Where does the data and its associated truth/metadata come from? While simulation has been around for decades, what’s new is a convergence in the maturity, realism, and availability of relatively simple-to-use tools and content/assets for individuals who are not computer graphics, physics, nor gaming experts. Companies like Epic Games, Google, Microsoft, Meta, OpenAI, Apple, NVIDIA, IBM, Tesla, Scale AI, and others have taken this a step further and developed billion-dollar in-house solutions based on synthetic data-driven AI. The 2023 IEEE AIPR Workshop will explore AI/ML/DL in synthetic, augmented, and natural datasets.

In addition to papers on regular AIPR topics in applied imagery, as they pertain to computer vision, imaging, and pattern recognition, the Workshop Committee invites papers focused on, but not limited to, the following:
- Theories, frameworks, and workflows to generate synthetic, augmented, and/or natural datasets;
- Novel solutions for interfacing synthetic, augmented, and/or natural datasets;
- Digital twins, Omniverse, Metaverse, etc. for representation, modeling, simulation
- Neural radiance fields (NeRFs) for virtual environments
- AI/ML/DL evaluated on synthetic, augmented, and/or natural datasets
- Synthetic or augmented approaches for training AI/ML/DL algorithms
- Synthetic data sets for training AI/ML/DL in biomedical imaging, medicine, healthcare, life science
- Trustworthy and safe medical AI
- Zero shot, few shot learning, domain generalization using synthetic datasets
- Verification and validation (V&V); uncertainty quantification; responsible open source AI, trustworthy AI
- Generative techniques, large language models, transformers
- Simulation of multispectral imagery or non-traditional sensor data
- Fusion of multisource imagery data (SAR, optical, thermal, and LiDAR)
- Fusion of synthetic, augmented, and/or real data at the data, signal, feature, and/or algorithm level
- Transferring simulated/augmented datasets and/or AI/ML/DL models to real data
- Approaches for generating accurate and dense truth and metadata
- Controlled synthetic/augmented studies that go beyond what is practical or possible in the real-world
- Explainable AI (XAI) and evaluating/characterization/understanding of AI/ML/DL algorithms
- Applications in remote sensing including agriculture, climate security, arctic navigability, etc.
- Scalable approaches for computer vision, change detection, structure from motion, merging real objects with virtual worlds, etc.
- Ways to produce, structure, store, and format synthetic/augmented data for AI/ML/DL;
- Synthetic/augmented/natural datasets for autonomous vehicles, clinical medical imaging;
- Human-in-the-loop (HITL) or human-over-the-loop (HOTL) simulation;
- Closing the loop and inverse design
-

Deadline for abstracts: July 17th, 2023. The Workshop will include oral and poster presentations, several keynote talks that provide in-depth overviews of the fields, and a special session on the theme topic. Accepted papers will be submitted for inclusion into IEEE Xplore subject to meeting IEEE Xplore's scope and quality requirements. AIPR 2023, the 52nd annual workshop, is sponsored by the IEEE Computer Society Technical Committee on Pattern Analysis and Machine Intelligence, and organized by the AIPR Workshop Committee with generous support from sponsors.