American Hand Signs Dataset
Description
This dataset consists of 17,600 labeled images of American Sign Language hand signs, including representations of the alphabets (A to Z) and the "I love you" gesture. The images are of size 300x300 pixels and contain various background and lighting conditions to ensure robustness. Each alphabet sign contains approximately 600 images. The dataset was generated using MediaPipe for hand gesture detection and was trained using TensorFlow and Keras, achieving high accuracy on the dataset. The dataset is intended for use in the development and evaluation of machine learning models for hand gesture recognition. It is suitable for applications in sign language recognition, human-computer interaction, and other related fields. The variety in background and lighting makes it useful for real-world applications, ensuring that models trained on this dataset can generalize well to diverse environments. Keywords: American Sign Language, Hand Gesture Recognition, MediaPipe, TensorFlow, Keras, Image Dataset, Machine Learning, Sign Language Recognition, Computer Vision.
Files
Steps to reproduce
1. Set up the Environment: Install the necessary Python libraries for hand gesture detection and image processing by running: pip install cv2 cvzone numpy 2.Run the Data Collection Script: Use the following Python script to collect hand gesture images. The script uses the webcam to capture images of hand signs and saves them into a specified folder. The user will be prompted to input the gesture (e.g., A, B, or "I Love You"), and the script will store images in the corresponding folder (e.g., Data/A, Data/B). Press the 's' key to save an image and 'q' to stop the data collection process.(code file is given dataCollection.py) 3.Folder Structure: The images are saved in a folder named after the gesture, for example, Data/A/, Data/B/, etc. Each folder contains 600 images for that specific hand sign. 4.Keyboard Shortcuts: Press 's' to save an image of the detected gesture. Press 'q' to stop the data collection process. 5.Dataset Organization: Each hand gesture (e.g., "A", "B", or "I Love You") is saved in its respective folder. The images are saved with filenames containing the gesture name and a timestamp to prevent overwriting.
Institutions
- Vishwakarma Institute of Information Technology