Nina S. T. Hirata --- DCC-IME/USP

My main research interests lie in the intersection of two subject fields: image processing/analysis and machine learning.

Research projects / interests

Machine learning Understanding and applying machine learning
    Machine learning techniques and methods are core elements in the projects below.
    Topics covered: CNN, feature extraction and selection, ensembles, classifier combination, ...
TRIOSLib Learning image operators
    Designing image operators to perform an image processing task often requires specific knowledge and practical experience in the field. Achieving satisfactory results may require considerable trial and error effort. The main question under investigation in this project is how to learn an image transformation from samples of input-output image pairs.

    Learning image operators from training samples

    Related (past) project: Combinação de Classificadores
    FAPESP (01/July/2005 to 30/june/2007)

SegmentIt Image segmentation
    Segmentation of images may be effectively accomplished through user interaction. Developing new algorithms and easy interaction ways to reduce the required interaction effort is one of the main goals of this project.

    SegmentIt is an Interactive Image Segmentation Tool written in Java, developed by B. Klava, that allows handling image partition hierarchies as well as markers.

    See related publications at B. Klava's page ...
ExpressMath ExpressMath - Recognition of handwritten mathematical expressions
    Handwriting recognition is an interesting and challenging problem. There are many applications that could benefit from computational systems with handwriting recognition capabilities. This project is concerned with recognition of online handwritten math expressions.

    See more details here
Document images Document image processing
    Many interesting problems that can be addressed using machine learning and graph modeling methods.
Astronomy Analysis of astronomical data
    This is a project within NAP-LabCosmos, in collaboration with researchers at IAG-USP and IF-USP. The main tasks are related to image/data processing and analysis, mainly those coming from large-scale astronomical sky surveys based on optical telescopes.
Plankton Plankton image processing and analysis
Past/Frozen projects
    Environment for exploratory image analysis: VisionDataset

    Quantification of angiogenesis: in collaboration with Prof. Lígia Ferreira Gomes (Faculdade de Ciências Farmacêuticas, Departamento de Análises Clínicas e Toxicológicas, USP)

    QR code detection.
    FastQR at sourceforge

    GPR data classification
    (matéria na Agência Universitária de Notícias)

    Detection of Schistosome eggs