Embedded Files
Publications
Journal Articles
Chakraborty, P., Rakib Mia, M., Sumon, H. K., Sarker, A., Imtiaz, A., Mahbubur Rahman, M., ... Choudhury, T. (2022). Recognize Meaningful Words and Idioms from the Images Based on OCR Tesseract Engine and NLTK. In Pattern Recognition and Data Analysis with Applications (pp. 297-310).
Singapore: Springer Nature Singapore.
DOI: 10.1007/978-981-19-1520-823
Study the complete work from here....
Abstract: OCR means optical character recognition, which is a text extraction technology that works with photos, scanned data, and PDF documents. By extracting text data, OCR systems typically convert non-editable, nonsearchable documents into editable, searchable files. As a result, information finding and identification from digitized files is simplified. R bindings are provided by the Tesseract package. Tesseract is a strong optical character recognition (OCR) engine with over 100 languages supported. The engine is highly customizable, allowing you to fine-tune the detection algorithms to achieve the best possible results. With the help of Tesseract OCR technology, a method for extracting texts from photos was created. Any image can be used as input for the proposed OCR system, which converts it into a searchable text document. Furthermore, this system can search for words within the generated text and display the Bengali meaning terms. It finds the words and lines first, then identifies the words, then the static character classifier classifies the character, then does analysis, and finally an adaptive classifier. It is a framework that also includes a natural language processing approach for classifying commonly used terms with Bangla meanings from the output text, in addition to OCR.
Extended Abstract
Khatun, M., Rakib Mia, M., Tafhim, S. N., Chakraborty, P., Hasan, M. R., Ahmed, T., Sultana, M., Sarker, A.. Image Classification of Face Mask and Face Shield Using CNN In the Era of COVID-19. In IEEE CSBDC Winter Symposium 2020. IEEE Computer Society Bangladesh Chapter.
Detection of the face shield and non-face shield images by using various object detection algorithms and comparing those results.
The algorithms are Faster R-CNN, YOLOV5, and MobileNet SSD.
The main intention of our study is to identify the safety equipment of people in order to prevent the spread of COVID-19
Prior Publication
Khatun, M., Rakib Mia, M.,. Object Detection using Deep Learning: HOG, R-CNN, Fast R-CNN, Faster R-CNN, YOLO, SSD, and SPP-net using Own Customized Dataset.
Object detection is a prominent machine learning and image processing application. In this era of COVID-19, ensuring safety is mandatory. Several studies have been done to detect face masks from images and videos. The concept of detecting face shields is almost new. There is no study found for the detection of face shields yet. In this paper, we try to detect face shields from different images using different deep-learning algorithms for object detection. Our study shows that MobileNet SSD gives the best output for detecting face shields.
Bangladeshi Police Dress Identification Using Deep Learning, (University Research Project-2022)
Abstract- The technique of recognizing an item or feature in an image or video is known as image recognition. It is employed in several contexts, including flaw identification, imaging in the medical field, and security monitoring. Most vision-based AI systems and applications today are focused on object recognition. Object recognition is crucial for clear determination, which is helpful in security, transportation, healthcare, and military use cases object detection in retail. Due to the large range of dress designs and the intricate nature of their commonalities, it is difficult to precisely categorize police uniforms. In Bangladesh, uniforms for the police and security forces come in a variety of hues. This makes it difficult to improve performance using a simple classification system. However, using this strategy will ensure that you may conduct study in this exciting field. Consequently, here is the technique for identifying Bangladeshi police uniforms from a picture was developed. In order to achieve these goals, we first constructed the CNN model for predicting Bangla digits since CNN is the most intelligent object prediction model. After a thorough examination, the suggested system attained a better accuracy on our own customized dataset.
Page updated
Google Sites
Report abuse