.jpg)
Hi, I'm Amal Joseph
PhD CSE @ Center for Visual Information Technology (CVIT), IIIT Hyderabad advised by Prof Ravi Kiran
Ex-Machine Learning Engineer @ Arcesium
Checkout my project course on "implementing UNET for Image Segmentation from scratch" at Educative.
Publications
EpiSAM: Character Segmentation in Challenging Stone Inscriptions
Arnav Sharma, Pratyush Jena, Amal Joseph, Ravi Kiran Sarvadevabhatla
In International Conference on Document Analysis and Recognition (ICDAR) , 2026
EpiSAM is a prompt-guided transformer framework for stone inscription character segmentation that uses neighboring character context to improve detection and boundary accuracy in challenging epigraphic images.
Unveiling Text in Challenging Stone Inscriptions: A Character-Context-Aware Patching Strategy for Binarization
Pratyush Jena, Amal Joseph, Arnav Sharma, Ravi Kiran Sarvadevabhatla
In Indian Conference on Computer Vision, Graphics, and Image Processing (ICVGIP) , 2025
A novel Character-Context-Adaptive patching strategy for Binarization and a pixel-precise challenging Indic stone inscription dataset.
.png)
LineTR: Unified Text Line Segmentation for Challenging Palm Leaf Manuscripts
Vaibhav Agrawal, Niharika Vadlamudi, Muhammad Waseem, Amal Joseph, Sreenya Chitluri, Ravi Kiran Sarvadevabhatla
In International Conference on Pattern Recognition (ICPR) , 2024
Re-imagining text line segmentation in challenging documents. Instead of a pixel-based segmentation paradigm, LineTR uses a parametric representation of a line, leveraging its inductive priors. Introduces a novel context-adaptive patching mechanism for zero-shot generalizability.