Latest Issue
Evaluation of Dispersive Soil Characteristics in Cambodia
Published: October 31,2026The Special Character of Angkor Sand and the Preservation of the Central Tower of Bayon Temple
Published: October 31,2026Case Studied of Fundamental of soil properties and their correlation of soil in Battambang Provinces
Published: October 31,2026Point Cloud Data from Different Measurement Methods in Road Embankments
Published: October 31,2026Basic Study on Problem Soils Encountered in Construction Works in Developing Countries
Published: October 31,2026Development of SPT and DCPT data Correlations for Foundation Design in Phnom Penh Capital City of Cambodia.
Published: October 31,2026Soil Stabilization using Cementless Geopolymer
Published: October 31,2026Effect of Root Orientation and Decay on Soil Shear Strength
Published: October 31,2026Word Spotting on Khmer Printed Documents
-
1. Department of Information and Communication Engineering,, Institute of Technology of Cambodia, Russian Federation Blvd., P.O. Box 86, Phnom Penh, Cambodia
Received: August 08,2024 / Revised: September 10,2024 / / Accepted: September 17,2024 / Available online: August 30,2025
Word spotting in Khmer printed documents presents a unique challenge due to the complexities of the Khmer script and the vast array of font styles employed. The scarcity of large, publicly available datasets further complicates this task. This work proposes a two-module approach for achieving accurate and efficient word spotting in Khmer documents. Separate datasets are utilized for text detection and recognition. The first module employs the state-of-the-art YOLOv8 model on a dataset of 10,050 text samples. The model's performance is evaluated using the F1 score, a metric that balances precision and recall in locating text. The second module leverages the fine-tuned Transformer-based TrOCR model for recognition, trained on 22,567 labeled words, with recognition accuracy measured by the Character Error Rate (CER). The first module achieves an impressive F1 score of 0.987 in locating Khmer words within documents. The second module's TrOCR model results in a CER of 8.41%. By overcoming script and font challenges through focused datasets and advanced models, this approach demonstrates potential for improving document processing and information retrieval for the Khmer language.
