Research Projects
2019
This is a first course on Bioimage Understanding. The course has emphasis on the following three introductory areas:
- Lego Blocks of Visual Recognition Knowing the fundamental visual recognition tasks, being able to perform such task and convert the biological problem statements into one or more of the fundamental tasks
- Open Source Software Ecosystem Getting familiar with the widely used image analysis tools
- Coding workshop This is a project oriented module to help students learn coding for image analysis
2017
In this work we propose a fundamentally different representation for image templates in the form of multidimesional tensors that looks beyond the histogram features of HOG by aggressively capturing local image geometry. Robust to noise and signal perturbation, the proposed tensor representation of templates yields excellent localization performance in unconventional and difficult scenarios (e.g., low resolution, noisy image or video) where traditional HOG features fail to perform well. Moreover, owing to signal processing techniques, tensors are amenable a to rich set of tools that make object detection fast, efficient and scalable. Building on these premises we have proposed a maximum margin formulation following a relatively simple and fast training phase, to detect pedestrians in challenging videos of infrared, thermal images.
Sujoy Kumar Biswas, Peyman Milanfar, Linear Support Tensor Machine: Pedestrian Detection in Thermal Infrared Images , IEEE Transactions on Image Processing, 2017
2016
One shot object detection involves searching a given query object in a target image. For rapid detection we need low dimensional but locally discriminative features. How to derive such embedding? How to achieve extremely fast processing of an image/video frame, and at the same time compute detection scores at every pixel of the image? In other words, how to overcome the shortcoming of expensive sliding window based detection? How to decide detection threshold in absence of prior knowledge/training? The objective of this project is to address such concerns.
Sujoy Kumar Biswas, Peyman Milanfar, Laplacian Object : One Shot Object Detection with Locality Preserving Projections, International Conference in Image Processing (ICIP), 2014 (Accepted for Oral Presentation) (PDF)
Sujoy Kumar Biswas, Peyman Milanfar, One Shot Detection with Laplacian Object and Fast Matrix Cosine Similarity, IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015 (PDF)
2011
We propose a generative model for constructing an efficient set of distinctive textures for recognizing architectural distortion in digital mammograms. In the first layer of the proposed two-layer architecture, the mammogram is analyzed by a multi-scale, oriented filter bank to form texture descriptor of vectorized filter responses. Our model presumes that every mammogram can be characterized by a “bag of primitive texture patterns” and the set of textural primitives (or textons) is represented by a mixture of Gaussians which builds up the second layer of the proposed model. The observed textural descriptor in the first layer is assumed to be a stochastic realization of one (hard mapping) or more (soft mapping) textural primitive(s) from the second layer. The results obtained on two publicly available datasets, namely mini-MIAS and DDSM, demonstrate the efficacy of the proposed approach.
Sujoy Kumar Biswas, Dipti Prasad Mukherjee, Recognizing Architectural Distortion in Mammogram: A Multi-Scale Texture Modeling Approach with GMM, IEEE Transactions on Biomedical Engineering, 2011