Drawings and abstract Imagery: Representation and Analysis (DIRA) Workshop

Tel Aviv, Israel 2022

sites.google.com/view/eccv-dira/home

In conjunction with ECCV 2022

Illustrations, drawings, technical diagrams, and charts can communicate difficult concepts very quickly to humans. This type of visual information is prevalent in technical, scientific, and scholarly publications; as well as instruction manuals, textbooks, and infographics in popular media. However, many open challenges in computer vision for automatically processing this visual data remain. These abstract images have very different properties from natural images; for example, informative pixels are sparse, while the shape and spatial information are more important than color, texture, and shading which dominate natural images. While deep learning methods are quite successful in object detection and content-based retrieval in natural images; it is not clear when these approaches will transfer successfully to these much sparser abstract drawings. 

The DIRA workshop aims to bring together researchers from both industry and academia who are working in diverse areas of research in the representation, analysis, and applications of abstract images, such as illustrations, drawings, technical diagrams, charts, and plots. The goal of the workshop is to explore and highlight technical challenges, insights, and solutions that exist in representation, information extraction, semantic understanding, content-based retrieval, question-answering, and various other topics pertaining to the representation and analysis of abstract images, which has traditionally lagged behind progress in their natural image counterparts. 

The DIRA workshop will also host a data competition focused on abstract drawing retrieval on the DeepPatent dataset to encourage increased research and engineering activity in this field.

Invited Speakers

• Niloy Mitra, University of College London
• Yulia Gryaditskaya, University of Surrey
• Anjali Adukia, University of Chicago
• Changjian Li, INRIA Sophia-Antipolis

Topics of Interest

The DIRA workshop is currently soliciting papers describing new work in the following subject areas, as well as any other subject relevant to automated understanding of sparse images

1. Image Analysis
2. Image Segmentation
3. Semantic Image Understanding
4. Geometric Deep Learning
5. Sparse Machine Learning Models
6. Information Retrieval for Technical Documents
7. Knowledge Graph Construction
8. Optical Character Recognition
9. Feature Extraction
10. Multimodal Machine Learning
11. Plagiarism Detection

Additionally, the workshop seeks dataset papers describing new datasets in the subject areas of:

1. Technical Diagrams
2. Illustrations
3. Plots and Charts
4. Scientific Reports
5. Patents

Important dates

Paper submission deadline (Full papers): July 1, 2022 

Notification to Authors: August 8, 2022

Camera-ready submission deadline: August 22, 2022

Organizers

• Diane Oyen, Los Alamos National Laboratory
• Kushal Kafle, Adobe Research
• Pradyumna Reddy, University College London
• Michal Kucer, Los Alamos National Laboratory
• Cory B. Scott, Colorado College