This dataset includes 51 unique sign gestures for Bangla Sign Language words. The signs were performed by two signers. Each signer performed all the signs ten times, making the total number of videos in this dataset 1020 (51*2*10 videos).In the dataset folder, there are two folders titled "MahidulDataset" and "ShahidulDataset". Each folder contains signs from a particular signer. Each folder has 51 sub-folders, and each sub-folder has 10 signs. The subfolders are named "_", for example, "Shahidul_008". The association of sign labels to sign words is shown in the "sign_labels.xlsx" file.

Bangla Sign Language

## Overview

Bangla Sign Language is a dataset for object detection tasks - it contains ChandraBindu annotations for 2,412 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

BdSL47 is the first open-access complete dataset in Bangla Sign Language that contains hand signs from both 10 sign digits (from sign ০ to sign ৯) and 37 sign alphabet (from sign অ to sign ँ).

Dataset summary :

• 100 RGB images per sign (total 47 signs) from each of 10 users
• Total input images : 100×47×10 = 47000
• Input images are processed via MediaPipe, which provided
  • an output image with hand key-points being detected
  • 3D coordinate values of 21 predefined key-points
  • Total 63 coordinate values for each sample
• The values are stored in csv files
• 1 CSV file contains values from 100 samples of 1 sign from 1 user
• Total CSV files : 47×10 = 470

The dataset has been made public for further research purposes. It is also available upon request here.

The Bangla Sign Language Detection System is an innovative computer vision project aimed at developing a robust and efficient system to detect Bangla sign language gestures, specifically focusing on digits and alphabets. This project utilizes the power of YOLOv5, a state-of-the-art object detection algorithm, to achieve high accuracy and real-time performance.

Objective:

The primary objective of this project is to create a computer vision model that can accurately recognize and interpret Bangla sign language gestures for digits and alphabets. The system aims to facilitate communication for individuals with hearing and speech impairments, enabling them to interact effectively with others using sign language.

Methodology:

The project employs the YOLOv5 architecture, which is renowned for its speed, accuracy, and ease of implementation. YOLO (You Only Look Once) is a single-shot object detection model that simultaneously predicts bounding boxes and class probabilities for each object in an image. The model's architecture allows it to achieve real-time performance on various hardware platforms.

Performance Metrics:

The Bangla Sign Language Detection System using YOLOv5 has achieved remarkable performance in terms of accuracy and precision, with the following metrics:

96.4% mAP (Mean Average Precision): This metric measures the average precision across all classes and detection confidence thresholds, reflecting the overall accuracy of the model's predictions.

94.8% Precision: Precision measures the proportion of true positive detections out of all positive detections, indicating the model's ability to avoid false positives.

93.4% Recall: Recall, also known as sensitivity, measures the proportion of true positive detections out of all ground truth positive instances, indicating the model's ability to avoid false negatives.

Dataset:

To train the YOLOv5 model, a comprehensive and diverse dataset of Bangla sign language gestures for digits and alphabets was collected and annotated. The dataset includes various hand orientations, lighting conditions, and backgrounds, ensuring the model's robustness in real-world scenarios.

Implementation:

The project's implementation involves the following key steps:

Data Preprocessing: Cleaning, resizing, and augmenting the dataset to ensure diversity and mitigate overfitting.

Model Training: Utilizing YOLOv5 and training the model on the annotated dataset, tuning hyperparameters for optimal performance.

Model Evaluation: Validating the trained model using a test set to measure performance metrics such as mAP, precision, and recall.

Inference: Deploying the trained model in real-time applications to detect Bangla sign language gestures.

Expected Impact:

The Bangla Sign Language Detection System using YOLOv5 has the potential to make a significant impact on the lives of individuals with hearing and speech impairments in Bangladesh. By accurately recognizing and interpreting sign language gestures, the system will facilitate smoother communication between the hearing-impaired community and the general population, fostering inclusivity and understanding.

Future Scope:

The project opens doors for further research and development in the domain of sign language detection. Future improvements may involve expanding the system to recognize more complex sign language expressions, incorporating finger-spelling, or integrating natural language processing for real-time translation of sign language into text or speech.

The Bangla Sign Language Detection System using YOLOv5 stands as a testament to the potential of computer vision technology to bridge communication gaps and empower individuals with diverse abilities. With its impressive accuracy and precision, the system paves the way for more accessible and inclusive societies.

Bangla sign language (BdSL) is a complete and independent natural sign language with its own linguistic characteristics. While there exists video datasets for well-known sign languages, there is currently no available dataset for word-level BdSL. In this study, we present a video-based word-level dataset for Bangla sign language, called SignBD-Word, consisting of 6000 sign videos representing 200 unique words. The dataset includes full and upper-body views of the signers, along with 2D body pose information. This dataset can also be used as a benchmark for testing sign video classification algorithms.

Official Train Test Spllit (for both RGB and bodypose) can be found from the following link:
https://sites.google.com/view/signbd-word/dataset

This dataset is part of the following paper:
A. Sams, A. H. Akash and S. M. M. Rahman, "SignBD-Word: Video-Based Bangla Word-Level Sign Language and Pose Translation," 2023 14th International Conference on Computing Communication and Networking Technologies (ICCCNT), Delhi, India, 2023, pp. 1-7, doi: 10.1109/ICCCNT56998.2023.10306914.

Download the corresponding paper from this link:
https://asnsams.github.io/Publications.html

Bangla Sign Language

## Overview

Bangla Sign Language is a dataset for object detection tasks - it contains Numbers Letters annotations for 2,180 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [MIT license](https://creativecommons.org/licenses/MIT).

Bangla Sign Language Dataset using Depth Information

This is a Bangla Sign Language Dataset, using MediaPipe framework, which accurately detects the hand & 21 hand key-points from a raw RGB image, and stores the co-ordinate values of these key-points. After collecting 47000 such raw image files for 47 signs (100 files per sign per user) and generating 47000 corresponding output image files applying MediaPipe, the co-ordinate values of these key-points are stored in a .csv files. This dataset contains 470 such .csv files (collected from 10 users for 47 signs in total). After generating the dataset, we have also done the classification, using different classifiers, such as KNN, SVM, RFC, DTC, Neural Networks etc. Accuracies for different classifiers are yielded in the classification code (in code section).

The KU-BdSL refers to a Bengali sign language dataset, which includes three variants of the data. The variants are - (i) Uni-scale Sign Language Dataset (USLD), (ii) Multi-scale Sign Language Dataset (MSLD), and (iii) Annotated Multi-scale Sign Language Dataset (AMSLD). The dataset consists of images representing single-hand gestures for BdSL alphabets. Several smartphones are taken into account to capture images from 39 participants (30 males and 9 females). These 39 participants associated with the dataset creation have not offered any financial benefit. Each version includes 30 classes that resemble the 38 consonants ('shoroborno') of Bengali alphabets. There is a total of 1,500 images in jpg format in each variant. The images are captured on flat surfaces at different times of the day to vary the brightness and contrast. Class names are Unicode values corresponding to the Bengali alphabets for USLD and MSLD.

Folder Names: 2433 -> ‘Chandra Bindu’ 2434 -> ‘Anusshar’ 2435 -> ‘Bisharga’ 2453 -> ‘Ka’ 2454 -> ‘Kha’ 2455 -> ‘Ga’ 2456 -> ‘Gha’ 2457 -> ‘Uo’ 2458 -> ‘Ca’ 2459 -> ‘Cha’ 2460-2479 -> ‘Borgio Ja/Anta Ja’ 2461 -> ‘Jha’ 2462 -> ‘Yo’ 2463 -> ‘Ta’ 2464 -> ‘Tha’ 2465 -> ‘Da’ 2466 -> ‘Dha’ 2467-2472 -> ‘Murdha Na/Donto Na’ 2468-2510 -> ‘ta/Khanda ta’ 2469 -> ‘tha’ 2470 -> ‘da’ 2471 -> ‘dha’ 2474 -> ‘pa’ 2475 -> ‘fa’ 2476-2477 -> ‘Ba/Bha’ 2478 -> ‘Ma’ 2480-2524-2525 -> ‘Ba-y Ra/Da-y Ra/Dha-y Ra’ 2482 -> ‘La’ 2486-2488-2487 -> ‘Talobbo sha/Danta sa/Murdha Sha’ 2489 -> ‘Ha’

USLD: USLD has a unique size for all the images that is 512*512 pixels. The intended hand position is placed in the middle of the majority of cases in this dataset. MSLD: The raw images are stored in MSLD so that researchers can make changes to the dataset. The use of various smartphones yields us a wide variety of image sizes. AMSLD: AMSLD has multi-scale annotated data, which is suitable for tasks like localization and classification. From many annotation formats, the YOLO DarkNet annotation has been selected. Each image has an annotation text file containing five numbers separated by white space. The initial number is an integer, and the rest are floating numbers. The first number of the file indicates the class ID corresponding to the label of that image. Class IDs are mapped in a separate text file named 'obj.names'. The second and third values are the beginning normalized coordinates, while the fourth and fifth define the bounding box's normalized width and height.

This dataset is supported by Research and Innovation Center, Khulna University, Khulna-9208, Bangladesh and all the data from this dataset is free to download, modify, and use. The previous version (Version 1) of this dataset contains the oral permission of the volunteers, and the rest versions have written consent of the participants. Therefore, we encourage researchers to use these versions (Version 2 or Version 3 or Version 4) for research objective.

Bangla Sign Language Digits and Characters. - Digits: 1000 images - Characters: 1800 images

The dataset contains 47000 RGB input images of 47 signs (10 digits, 37 letters) of Bangla Sign Language. The images have been processed via MediaPipe framework, which is designed to detect predefined 21 hand key-points from a sample and provide normalized x & y coordinate values and an estimated depth value. The 3D coordinate values were stored in .csv files (1 file contains information of 100 image sample of the same sign). The dataset contains 470 .csv files in total, and 47000 corresponding output images with hand key-points being detected.

postures

Context

Bangla Sign Digits Recognition

Content

Dataset includes images of sign digits from 0 to 9 in Bangla Sign Language, corresponding output images showing the detected hand key-points using Mediapipe, and corresponding .csv file for each sign, that contains the co-ordinates information of those points. There are 1000 input images in total, 100 from each sign.

English Sign Language Detection with hand gestures

Continuous Bangla Sign Language Dataset

👐 A Rich Multimodal Dataset for Continuous Bangla Sign Language Understanding

  📁 Dataset Overview

This dataset is a large-scale, curated resource for research in Continuous Bangla Sign Language (BdSL) recognition and translation. It provides a comprehensive multimodal collection that includes:

🎥 Video Samples: Real-life continuous sign language performances. ✍️ Gloss Sentences: Base-form Bangla words representing the signed… See the full description on the dataset page: https://huggingface.co/datasets/banglagov/Ban-Sign-Sent-9K-V1.

The BdSL47 is the original dataset constructed under the supervision of the Systems and Software Lab (SSL), Department of Computer Science and Engineering (CSE), Islamic University of Technology (IUT) and Department of Computer Science and Engineering (CSE), United International University (UIU), Dhaka, Bangladesh. The ownership of this dataset belongs to the authors and they have administered the dataset collection processes by taking informed consent from the participants/singers of the Bangla sign Alphabets and Digits. The dataset was constructed by collecting Bangla alphabet signs from 10 signers in a controlled setting. We employed varying factors of the users like age, gender, hand shape, and skin color. We have also incorporated different challenges while collecting data like scaling, translation, hand rotation, hand orientation, lighting ambiance, and background. We have collected webcam images of 47 signs (10 Bangla Sign Digits and 37 Bangla Sign Alphabet) and resized them to 640x480. For each of the RGB image samples, we have detected the hand key points via the MediaPipe library. Then we generate CSV files containing x, y, and depth coordinate values of 21 hand key points extracted from these image samples and constitute 21×3, or 63 input features.

The dataset contains 47000 RGB input images of 47 signs (10 digits, 37 letters) of Bangla Sign Language. The images have been processed via the MediaPipe framework, which is designed to detect predefined 21 hand key points from a sample and provide normalized x & y coordinate values and an estimated depth value. The 3D coordinate values were stored in .csv files (1 file contains information on 100 image samples of the same sign). The dataset contains 470 .csv files in total.

There are two folders named as “Bangla Sign Language Dataset - Sign Alphabets” and “Bangla Sign Language Dataset - Sign Digits”. Under each folder, user-wise folders are given that contain sign images (input images (raw images in jpg format), and CSV files (normalized 3D coordinates of 21 hand keypoints with corresponding class labels). All the files of around 1.12GB are available for direct download using the ‘Download All’ option by going to the doi: 10.17632/pbb3w3f92y.3. The images and CSV files can be easily read or processed using Python or any other programming language (Python 3.10.3). Please cite our dataset if you have used it in your research following the format, “Rayeed, S M; Tuba, Sidratul Tamzida; Mahmud, Hasan; Mazumder, Mumtahin Habib Ullah; Hossain, Md. Saddam; Hasan, Md. Kamrul (2023), “BdSL47: A Complete Depth-based Bangla Sign Alphabet and Digit Dataset”, Mendeley Data, V3, doi: 10.17632/pbb3w3f92y.3”. The files associated with this dataset are licensed under a Creative Commons Attribution 4.0 International license. The link to the code can be found here: https://github.com/SMRayeed/BdSL47-Recognition.

Bangla Signs Recognition

## Overview

Bangla Signs Recognition is a dataset for object detection tasks - it contains Signs ULUf annotations for 10,133 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

most previous works focus on American Sign Language detection

Bangla Road Sign

## Overview

Bangla Road Sign is a dataset for object detection tasks - it contains Objects annotations for 234 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

BanglaSignLabguage

## Overview

BanglaSignLabguage is a dataset for object detection tasks - it contains Language annotations for 4,013 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

BdSLW - 11 stands for Bangladeshi Sign Language Words dataset for 11 sign words. This dataset is a collection of 1105 images of 11 classes or categories. Each class of the dataset contains more than 77 images. The classes are labeled as the 11 BdSL daily useful common Sign Words. The classes are 'Bad', 'Beautiful',' Friend', 'Good',' House',' Me',' My', 'Request',' Skin',' Urine' and 'You'. The images are selected and processed focusing on only hand gestures, clear background, and brightness. The size of the images is 224 x 224 and the format of the images is RGB with high resolution. Images are taken from volunteer signers with full permission by smartphones. This dataset is greatly helpful for the Deaf and Dumb community and also researchers.