San Jose

Lubomir Bourdev
Senior Research Scientist

I am a researcher in the Advanced Technology Labs at Adobe interested in person and object recognition and high-level computer vision. I received my Ph.D. in Computer Science from U.C. Berkeley (advisor: Prof. Jitendra Malik), and my Bachelor's and Master's degrees from Brown University (coadvised by Prof. Andy van Dam and Prof. John Hughes.)

I have been a member of Adobe’s ATL since 1998.

projects and activities

Poselets — we are exploring novel computer vision parts, called poselets, in collaboration with U.C. Berkeley. Poselets are very effective in image and video understanding (specifically, object recognition, semantic segmentation, pose estimation, action and attribute classification, gender recognition). We have state-of-the-art results on some of these computer vision problems.
People Recognition — I have been working on user-assisted face recognition using temporal and image context. Our engine is used in Photoshop Elements 8.0+. (Joint work with Alex Parenteau and the Elements team)
Face Tagging — a Photoshop Elements 4.0 feature that finds the faces of people in one's photo album and allows the user to tag them faster and easier. (integrated into Photoshop Elements by Claire Schendel).
The Soft Cascade — a method for training face detectors with state-of-the-art accuracy and speed (joint work with Jonathan Brandt)
The Flattening Preview — an interactive transparency-focused preview tool used in Illustrator
Transparency Flattening — the technology to flatten transparent vector graphics into opague ones, which is used when printing from Illustrator, InDesign and Acrobat (joint work with Stephen Schiller)
The Symbolism Tools — a set of creative tools in Illustrator that make it easy to draw natural scenes, such as grass, hair, shading, clouds and pen & ink type of illustrations
AutoFill Engine — the technology behind Acrobat's form filling feature. It learns patterns of form entries and can extrapolate them to new forms (integrated into Acorbat by Alex Mohr)
Generic Image Library — an open-source C++ image library that abstracts away image representations from algorithms without compromising speed. It is accepted into the Boost libraries. (joint work with Hailin Jin).

publications

• Bourdev, L., Maji, S., and Malik, J. 2011. Describing People: A Poselet-Based Approach to Attribute Classification. In International Conference on Computer Vision (ICCV 2011).
• Hariharan, B., Arbelaez, P., Bourdev, L., Maji, S., and Malik, J. 2011. Semantic Contours Using Inverse Detectors. In International Conference on Computer Vision (ICCV 2011).
• Pongnumkul, S., Dontcheva, M., Li, W., Bourdev, L., Avidan, S., Wang, J., Cohen, M., Pause-and-play: Automatically Linking Screencast Video Tutorials with Applications, In ACM Symposium on User Interface Software and Technology (UIST 2011).
• Yang, F., Wang, J., Shechtman, E., Bourdev, L., Metaxas, D. 2011. Expression Flow for 3D-Aware Face Component Transfer. ACM Transactions on Graphics (Proceedings of SIGGRAPH 2011).
• Brox, T., Bourdev, L., Maji. S., and Malik, J. 2011. Object Segmentation by Alignment of Poselet Activations to Image Contours. In Conference on Computer Vision and Pattern Recognition (CVPR 2011).
• Maji, S., Bourdev, L. and Malik, J. 2011. Action Recognition From a Distributed Representation of Pose and Appearance. In Conference on Computer Vision and Pattern Recognition (CVPR 2011).
• Bourdev, L. 2011. Poselets and Their Applications in High-Level Computer Vision. Ph.D. Thesis, U.C.Berkeley.
• Bourdev, L., Maji, S., Brox, T. and Malik, J. 2010. Detecting People using Mutually Consistent Poselet Activations. In European Conference on Computer Vision (ECCV 2010).
• Bourdev, L. and Malik, J. 2009. Poselets: Body Part Detectors Trained Using 3D Human Pose Annotations. In International Conference on Computer Vision (ICCV 2009).
• Bourdev, L. and Järvi, J. 2009. Efficient run-time dispatching in generic programming with minimal code bloat. Science of Computer Programming.
• Bourdev, L. 2007. Generic image library. Software Developer's Journal, 42–52.
• Bourdev, L. and Järvi, J. 2006. Efficient run-time dispatching in generic programming with minimal code bloat. In Workshop of Library-Centric Software Design at OOPSLA'06, Portland Oregon, 15–24.
• Bourdev, L. and Brandt, J. 2005. Robust object detection via soft cascade. In 2005 Conference on Computer Vision and Pattern Recognition (CVPR 2005), 236–243.
• Kowalski, M. A., Markosian, L., Northrup, J. D., Bourdev, L., Barzel, R., Holden, L. S., and Hughes, J. F. 1999. Art-based rendering of fur, grass, and trees. ACM Transactions on Graphics (Proceedings of SIGGRAPH 1999), 433–438.
• Markosian, L., Kowalski, M. A., Trychin, S. J., Bourdev, L. D., Goldstein, D., and Hughes, J. F. 1997. Real-time nonphotorealistic rendering. ACM Transactions on Graphics (Proceedings of SIGGRAPH 1997), 415–420.

patents

• 7,966,566 — Reviewing and Editing Word Processing Documents
• 7,889,946 — Facilitating Computer-Assisted Tagging of Object Instances in Digital Images
• 7,827,485 — Previewing the Effects of Flattening Transparency
• 7,825,941 — Processing Illustration Artwork
• 7,818,741 — Method and System to Monitor Installation of a Software Program
• 7,813,557 — Tagging Detected Objects
• 7,813,526 — Normalizing Detected Objects
• 7,768,516 — Image Splitting to Use Multiple Execution Channels of a Graphics Processor to Perform an Operation on a Single Channel Input
• 7,738,680 — Detecting an Object within an Image by Incrementally Evaluating Subwindows in the Image in Parallel
• 7,730,043 — Incremental Batch-Mode Editing of Digital Media Objects
• 7,716,157 — Searching Images with Extracted Objects
• 7,706,577 — Exporting Extracted Faces
• 7,694,885 — Indicating a Tag with Visual Data
• 7,692,652 — Selectively Transforming Overlapping Illustration Artwork
• 7,636,450 — Displaying Detected Objects to Indicate Grouping
• 7,634,142 — Detecting Objects in an Image Using a Soft Cascade
• 7,616,780 — Method and Apparatus for Calibrating Sampling Operations for an Object Detection Process
• 7,587,101 — Facilitating Computer-Assisted Tagging of Object Instances in Digital Images
• 7,580,563 — Detection of Objects in an Image using Color Analysis
• 7,502,028 — Assigning Region Attributes in a Drawing
• 7,495,675 — Processing Illustration Artwork
• 7,440,587 — Method and apparatus for calibrating sampling operations for an object detection process
• 7,343,551 — Autocompleting Form Fields Based on Previously Entered Values
• 7,339,597 — Creating and Manipulating Related Vector Objects in an Image
• 7,262,782 — Selectively transforming overlapping illustration artwork
• 7,256,798 — Processing Complex Regions of Illustration Artwork
• 7,181,687 — Previewing the Effects of Flattening Transparency
• 7,123,269 — Operations on a Related Set of Vector Objects
• 7,102,651 — Hierarchical 2D Compositing with Blending Mode and Opacity at All Levels
• 6,894,704 — Processing Complex Regions of Illustration Artwork
• 6,859,553 — Flattening Images with Abstracted Objects
• 6,847,380 — Hierarchical 2D Compositing with Blending Mode and Opacity at All Levels
• 6,720,977 — Processing Illustration Artwork
• 6,515,675 — Processing Opaque Pieces of Illustration Artwork