Anil S. Baslamisli, PhD

Guest Researcher
Computer Vision Lab, Informatics Institute
University of Amsterdam

Email: a.s.baslamisli@uva.nl

[Google Scholar] [LinkedIn]

About Me

I obtained my PhD degree in 2021 at University of Amsterdam under the supervision of Prof. Theo Gevers. My doctoral study was fully funded by the EUHorizon2020 program within the Trimbot2020 project - the first ever autonomous gardening robot. My main research topic was intrinsic image decomposition, and my doctoral thesis is titled Physics-aware Learning of Intrinsic Images.

During my PhD, I also conducted research on semantic segmentation, object detection and recognition, instance segmentation, surface normal estimation, illumination estimation, optical flow, edge detection, and also traditional image processing, model-based machine learning, deep learning, and generative adversarial networks.

I received my Master's Degree in Data Engineering with distinction from Tampere University of Technology, Finland in 2016. I performed my thesis "Camera Sensor Invariant Auto White Balance Algorithm Weighting" under the supervison of Prof. Moncef Gabbouj and Dr. Jarno Nikkanen, in collaboration with Intel. During my studies, I also carried out an internship at Microsoft Finland, where I worked on image sharpness optimization and auto white balance problems.

My research interests include computer vision and pattern recognition, especially physics-based representations, invariant descriptors, general scene understanding, color image processing, and deep learning.

Publications

SAR Image Edge Detection: Review and Benchmark Experiments
Mees J. Meester and Anil S. Baslamisli
International Journal of Remote Sensing (IJRS), 2022. [Paper] [Project Page]

> Introducing the first-ever benchmark and experiments on SAR image edge detection methods.

Physics-based Shading Reconstruction for Intrinsic Image Decomposition
Anil S. Baslamisli, Yang Liu, Sezer Karaoglu and Theo Gevers
Computer Vision and Image Understanding (CVIU), 2021. [Paper] [Dataset - Coming Soon!..]

> Steering deep CNNs with physics-based descriptors achieves more accurate and robust albedo estimations.

ShadingNet: Image Intrinsics by Fine-Grained Shading Decomposition
Anil S. Baslamisli*, Partha Das*, Hoang-An Le, Sezer Karaoglu and Theo Gevers
International Journal of Computer Vision (IJCV), 2021. [Paper] [Project Page] [Dataset - Coming Soon!..]

> Further factorizing shading into different photometric effects improves albedo estimations.

Prior to Segment: Foreground Cues for Weakly Annotated Classes in Partially Supervised Instance Segmentation
David Biertimpel, Sindi Shkodrani, Anil S. Baslamisli and Nora Baka
IEEE International Conference of Computer Vision (ICCV), 2021. [Paper] [Project Page]

> Inducing an object mask prior enhances the generalization to the weakly annotated classes.

Invariant Descriptors for Intrinsic Reflectance Optimization
Anil S. Baslamisli and Theo Gevers
Journal of the Optical Society of America A (JOSA A), 2021. [Paper]

> Embedding physics-based descriptors as priors into optimization yields better albedo estimations.

Automatic Generation of Dense Non-rigid Optical Flow
Hoang-An Le, Tushar Nimbhorkar, Thomas Mensink, Anil S. Baslamisli, Sezer Karaoglu and Theo Gevers
Computer Vision and Image Understanding (CVIU), 2021. [Paper] [Project Page]

> Generating optical flow data from real videos is achievable, obviating the necessity of manual labeling.

Color Constancy by GANs: An Experimental Survey
Partha Das, Anil S. Baslamisli, Yang Liu, Sezer Karaoglu and Theo Gevers
arXiv preprint, 2018. [Paper]

> Utilizing GANs proves to be an effective tool in addressing the color constancy problem.

Three for One and One for Three: Flow, Segmentation, and Surface Normals
Hoang-An Le, Anil S. Baslamisli, Thomas Mensink and Theo Gevers
British Machine Vision Conference (BMVC), 2018. [Paper] [Project Page]

> Combining different sources of vision modality information jointly enhances each other.

Joint Learning of Intrinsic Images and Semantic Segmentation
Anil S. Baslamisli, Thomas T. Groenestege, Partha Das, Hoang-An Le, Sezer Karaoglu and Theo Gevers
European Conference on Computer Vision (ECCV), 2018. [Paper] [Project Page] [Dataset]

> Joint learning of intrinsics and semantics is beneficial for both tasks for natural scenes.

CNN based Learning using Reflection and Retinex Models for Intrinsic Image Decomposition
Anil S. Baslamisli, Hoang-An Le and Theo Gevers
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018. [Paper] [Project Page] [Dataset]

> Considering the physics-based reflection and Retinex models improves intrinsic image estimations.

Miscellaneous

I constantly provide reviews for several journals such as Transactions on Image Processing (TIP), International Journal of Computer Vision (IJCV), Transactions on Pattern Analysis and Machine Intelligence (TPAMI), The Journal of the Optical Society of America (JOSA), Optics Express, and Sensors.

I also provided reviews several times for the top conferences such as CVPR'19'20'21, BMVC'19'20, ICCV'19, AAAI'20'21, ECCV'20, ACCV'20'21 and WACV'21. However, I have stopped providing reviews for conferences since 2021.

Check out some of the amazing reviews we have received over the years from the top conferences and journals to get inspired for your research!

Check out my article on the Trimbot2020 project and our related research published on the UA Magazine!

Last updated on 26.02.2024.

Thanks to my bro Yahui Zhang for the template!