arXiv subjects

Parag Chaudhuri

Publications and source records attributed to Parag Chaudhuri.

10 recordsLinked to original sources

arXiv2025

DRISHTIKON: Visual Grounding at Multiple Granularities in Documents

Visual grounding in text-rich document images is a critical yet underexplored challenge for Document Intelligence and Visual Question Answering (VQA) systems. We present DRISHTIKON, a multi-granular and multi-block visual grounding framework designed to enhance interpretability and trust in VQA for complex, multilingual documents. Our approach integrates multilingual OCR, large language models, and a novel region matching algorithm to localize answer spans at the block, line, word, and point levels. We introduce the Multi-Granular Visual Grounding (MGVG) benchmark, a curated test set of diverse circular notifications from various sectors, each manually annotated with fine-grained, human-verified labels across multiple granularities. Extensive experiments show that our method achieves state-of-the-art grounding accuracy, with line-level granularity providing the best balance between precision and recall. Ablation studies further highlight the benefits of multi-block and multi-line reasoning. Comparative evaluations reveal that leading vision-language models struggle with precise localization, underscoring the effectiveness of our structured, alignment-based approach. Our findings pave the way for more robust and interpretable document understanding systems in real-world, text-centric scenarios with multi-granular grounding support. Code and dataset are made available for future research.

cs.CV↗

arXiv2025

SPRINT: Script-agnostic Structure Recognition in Tables

Table Structure Recognition (TSR) is vital for various downstream tasks like information retrieval, table reconstruction, and document understanding. While most state-of-the-art (SOTA) research predominantly focuses on TSR in English documents, the need for similar capabilities in other languages is evident, considering the global diversity of data. Moreover, creating substantial labeled data in non-English languages and training these SOTA models from scratch is costly and time-consuming. We propose TSR as a language-agnostic cell arrangement prediction and introduce SPRINT, Script-agnostic Structure Recognition in Tables. SPRINT uses recently introduced Optimized Table Structure Language (OTSL) sequences to predict table structures. We show that when coupled with a pre-trained table grid estimator, SPRINT can improve the overall tree edit distance-based similarity structure scores of tables even for non-English documents. We experimentally evaluate our performance across benchmark TSR datasets including PubTabNet, FinTabNet, and PubTables-1M. Our findings reveal that SPRINT not only matches SOTA models in performance on standard datasets but also demonstrates lower latency. Additionally, SPRINT excels in accurately identifying table structures in non-English documents, surpassing current leading models by showing an absolute average increase of 11.12%. We also present an algorithm for converting valid OTSL predictions into a widely used HTML-based table representation. To encourage further research, we release our code and Multilingual Scanned and Scene Table Structure Recognition Dataset, MUSTARD labeled with OTSL sequences for 1428 tables in thirteen languages encompassing several scripts at https://github.com/IITB-LEAP-OCR/SPRINT

cs.CV↗

arXiv2025

PLATTER: A Page-Level Handwritten Text Recognition System for Indic Scripts

In recent years, the field of Handwritten Text Recognition (HTR) has seen the emergence of various new models, each claiming to perform competitively better than the other in specific scenarios. However, making a fair comparison of these models is challenging due to inconsistent choices and diversity in test sets. Furthermore, recent advancements in HTR often fail to account for the diverse languages, especially Indic languages, likely due to the scarcity of relevant labeled datasets. Moreover, much of the previous work has focused primarily on character-level or word-level recognition, overlooking the crucial stage of Handwritten Text Detection (HTD) necessary for building a page-level end-to-end handwritten OCR pipeline. Through our paper, we address these gaps by making three pivotal contributions. Firstly, we present an end-to-end framework for Page-Level hAndwriTTen TExt Recognition (PLATTER) by treating it as a two-stage problem involving word-level HTD followed by HTR. This approach enables us to identify, assess, and address challenges in each stage independently. Secondly, we demonstrate the usage of PLATTER to measure the performance of our language-agnostic HTD model and present a consistent comparison of six trained HTR models on ten diverse Indic languages thereby encouraging consistent comparisons. Finally, we also release a Corpus of Handwritten Indic Scripts (CHIPS), a meticulously curated, page-level Indic handwritten OCR dataset labeled for both detection and recognition purposes. Additionally, we release our code and trained models, to encourage further contributions in this direction.

cs.CV↗

arXiv2024

Can AI Assistance Aid in the Grading of Handwritten Answer Sheets?

With recent advancements in artificial intelligence (AI), there has been growing interest in using state of the art (SOTA) AI solutions to provide assistance in grading handwritten answer sheets. While a few commercial products exist, the question of whether AI-assistance can actually reduce grading effort and time has not yet been carefully considered in published literature. This work introduces an AI-assisted grading pipeline. The pipeline first uses text detection to automatically detect question regions present in a question paper PDF. Next, it uses SOTA text detection methods to highlight important keywords present in the handwritten answer regions of scanned answer sheets to assist in the grading process. We then evaluate a prototype implementation of the AI-assisted grading pipeline deployed on an existing e-learning management platform. The evaluation involves a total of 5 different real-life examinations across 4 different courses at a reputed institute; it consists of a total of 42 questions, 17 graders, and 468 submissions. We log and analyze the grading time for each handwritten answer while using AI assistance and without it. Our evaluations have shown that, on average, the graders take 31% less time while grading a single response and 33% less grading time while grading a single answer sheet using AI assistance.

cs.AI↗

arXiv2024

Character Animation in AR: Character Animation in AR: a mobile application development study

Digital preservation of the cultural heritages is one of the major applications of various computer graphics and vision algorithms. The advancement in the AR/VR technologies is giving the cultural heritage preservation an interesting spin due to its immense visualization ability. The use of these technologies to digitally recreate heritage sites and art is becoming a popular trend. A project, called Indian Digital Heritage (IDH), for recreating the heritage site of Hampi, Karnataka during Vijaynagara empire ($1336$ - $1646$ CE) has been initiated by the Department of Science and Technology (DST) few years back. Immense work on surveying the site, collecting geographical and historic information about the life in Hampi, creating 3D models for buildings and people of Hampi and many other related tasks has been undertaken by various participants of this project. A major part of this project is to make tourists visiting Hampi visualize the life of people in ancient Hampi through any handy device. With such a requirement, the mobile AR based platform becomes a natural choice for developing any application for this purpose. We contributed to the project by developing an AR based mobile application to recreate a scene from Virupaksha Bazaar of Hampi with two components - author scene with augmented virtual contents at any scale, and visualize the same scene by reaching the physical location of augmentation. We develop an interactive application for the purpose of digitally recreating ancient Hampi. Though the focus of this work is not creating any static or dynamic content from scratch, it shows an interesting application of the content created in a real world scenario.

cs.HC↗

arXiv2024

Cube2Pipes : Investigating Hybrid Gameplay Using AR and a Tangible 3D Puzzle

We present our game, Cube2Pipes, as an attempt to investigate a unique gameplay design where we use a tangible 3D spatial puzzle, in the form of a 2X2 Rubik's Cube, as an interface to a tabletop mobile augmented reality (AR) game. The game interface adapts to user movement and interaction with both virtual and tangible elements via computer vision based tracking. This game can be seen as an instance of generic interactive hybrid systems as it involves interaction with both virtual and real, tangible elements. We present a thorough user evaluation about various aspects of the gameplay in order to answer the question as to whether hybrid gameplay involving both real and virtual interfaces and elements is more captivating and preferred by users, than standard (baseline) gameplay with only virtual elements. We use multiple industry standard user study questionnaires to try and answer this question. We also try to determine whether the game facilitates understanding of the spatial moves required to solve a Rubik's Cube, and the efficacy of a tangible puzzle interface to a tabletop AR game.

cs.HC↗

arXiv2024

TEXTRON: Weakly Supervised Multilingual Text Detection through Data Programming

Several recent deep learning (DL) based techniques perform considerably well on image-based multilingual text detection. However, their performance relies heavily on the availability and quality of training data. There are numerous types of page-level document images consisting of information in several modalities, languages, fonts, and layouts. This makes text detection a challenging problem in the field of computer vision (CV), especially for low-resource or handwritten languages. Furthermore, there is a scarcity of word-level labeled data for text detection, especially for multilingual settings and Indian scripts that incorporate both printed and handwritten text. Conventionally, Indian script text detection requires training a DL model on plenty of labeled data, but to the best of our knowledge, no relevant datasets are available. Manual annotation of such data requires a lot of time, effort, and expertise. In order to solve this problem, we propose TEXTRON, a Data Programming-based approach, where users can plug various text detection methods into a weak supervision-based learning framework. One can view this approach to multilingual text detection as an ensemble of different CV-based techniques and DL approaches. TEXTRON can leverage the predictions of DL models pre-trained on a significant amount of language data in conjunction with CV-based methods to improve text detection in other languages. We demonstrate that TEXTRON can improve the detection performance for documents written in Indian languages, despite the absence of corresponding labeled data. Further, through extensive experimentation, we show improvement brought about by our approach over the current State-of-the-art (SOTA) models, especially for handwritten Devanagari text. Code and dataset has been made available at https://github.com/IITB-LEAP-OCR/TEXTRON

cs.CV↗

arXiv2022

Physics-based Mesh Deformation with Haptic Feedback and Material Anisotropy

We present a physics-based framework to simulate porous, deformable materials and interactive tools with haptic feedback that can reshape it. In order to allow the material to be moulded non-homogeneously, we propose an algorithm to change the material properties of the object depending on its water content. We present a multi-resolution, multi-timescale simulation framework to enable stable visual and haptic feedback at interactive rates. We test our model for physical consistency, accuracy, interactivity and appeal through a user study and quantitative performance evaluation.

cs.GR↗

arXiv2022

Remeshing-Free Graph-Based Finite Element Method for Ductile and Brittle Fracture

Fracture produces new mesh fragments that introduce additional degrees of freedom in the system dynamics. Existing finite element method (FEM) based solutions suffer from an explosion in computational cost as the system matrix size increases. We solve this problem by presenting a graph-based FEM model for fracture simulation that is remeshing-free and easily scales to high-resolution meshes. Our algorithm models fracture on the graph induced in a volumetric mesh with tetrahedral elements. We relabel the edges of the graph using a computed damage variable to initialize and propagate fracture. We prove that non-linear, hyper-elastic strain energy is expressible entirely in terms of the edge lengths of the induced graph. This allows us to reformulate the system dynamics for the relabeled graph without changing the size of system dynamics matrix and thus prevents the computational cost from blowing up. The fractured surface has to be reconstructed explicitly only for visualization purposes. We simulate standard laboratory experiments from structural mechanics and compare the results with corresponding real-world experiments. We fracture objects made of a variety of brittle and ductile materials, and show that our technique offers stability and speed that is unmatched in current literature.

cs.GR↗

arXiv2017

A Deep Recurrent Framework for Cleaning Motion Capture Data

We present a deep, bidirectional, recurrent framework for cleaning noisy and incomplete motion capture data. It exploits temporal coherence and joint correlations to infer adaptive filters for each joint in each frame. A single model can be trained to denoise a heterogeneous mix of action types, under substantial amounts of noise. A signal that has both noise and gaps is preprocessed with a second bidirectional network that synthesizes missing frames from surrounding context. The approach handles a wide variety of noise types and long gaps, does not rely on knowledge of the noise distribution, and operates in a streaming setting. We validate our approach through extensive evaluations on noise both in joint angles and in joint positions, and show that it improves upon various alternatives.

cs.GR↗