The Sign Language Gesture Recognition project utilizes Python, OpenCV, and the powerful Random Forest algorithm. It captures and preprocesses video frames, extracting hand landmarks using Mediapipe. These landmarks are then used as input for the an small nerual network, trained on a labeled dataset of sign language gestures. The neural network learns patterns and correlations between landmarks and gestures, enabling accurate real-time recognition.
To install Sign Language Detection application, follow these steps:
- Clone the repository to your local machine:
git clone https://github.com/RaudelCasas1603/Monky-Detection-.git
- Navigate to the project directory:
cd Monky-Detection-
- Create a virtual environment (optional but recommended):
python3 -m venv env
- Activate the virtual environment:
- On macOS and Linux:
source env/bin/activate
- On Windows:
.\env\Scripts\activate
- On macOS and Linux:
- Install the package program:
pip install .
That's it! You have now successfully installed the Sign Language Detection application.
You can create your collection of data, firstly make sure that you already have installed the program in your python environment, then follow the next steps:
-
Open your terminal or command prompt.
-
Navigate to the directory where your eviroment is located. For example:
cd /path/to/your/env/sld
-
Activate the virtual environment if you have created one (optional):
source env/bin/activate # On macOS and Linux .\env\Scripts\activate # On Windows
-
Run the command with the collect-data command and specify the desired arguments:
sld collect-data -n 100 -d directory-to-store-data/ -s signs.json
- The -n 100 flag is optional and specifies that you want to generate 100 pictures per class or sign. By default, the program is programmed to take at least 1000 pictures.
- The -d directory-to-store-data/ flag is optional and sets the directory where the collected data will be stored. Replace directory-to-store-data/ with the actual directory path. If the directory doesn't exist, it will be created. By default, it uses the data/ directory.
- The -s signs.json flag is optional and specifies a JSON file with each sign to classify. By default, it uses the American alphabet for sign classification. Use this flag if you want to perform a custom sign classification.
-
The command will start collecting the data based on the provided arguments. It will generate pictures for each class and store them in the specified folder.
-
Once the data collection is completed, you will see the message "Data collection completed." printed in the terminal.
That's it! You have successfully created a data collection using the collect-data command. Adjust the arguments as needed to customize your data collection process.
Firstly make sure that you have collected your data, then to build your own dataset just follow the next steps:
- Open your terminal or command prompt.
- Navigate to the directory where your environment is located. For example:
cd /path/to/your/env/sld
- Activate the virtual environment if you have created one (optional):
source env/bin/activate # On macOS and Linux .\env\Scripts\activate # On Windows
- Run the command with the build-dataset command and specify the desired arguments:
sld build-dataset -f dataset.pickle -d data/
- -f dataset.pickle specifies the output filename of the built dataset. Replace dataset.pickle with the desired filename. By default, data.pickle is the output dataset filename.
- -d data/ sets the directory where the raw data is stored. Replace data/ with the actual directory. By default, the raw data is stored in the data/ directory.
- The command will start building the dataset based on the provided arguments. It will process the images from the specified data directory and generate the dataset file.
- Once the dataset is built, you will see the message "Dataset built." printed in the terminal.
blah blah not avaiable yet >:|
working ont eh new video
-
Computer vision engineer. (2023, January 26). Sign language detection with Python and Scikit Learn | Landmark detection | Computer vision tutorial [Video]. YouTube. https://www.youtube.com/watch?v=MJCSjXepaAM
-
Normalized Nerd. (2021, January 13). Decision Tree Classification Clearly Explained! [Video]. YouTube. https://www.youtube.com/watch?v=ZVR2Way4nwQ
-
Normalized Nerd. (2021b, April 21). Random Forest Algorithm Clearly Explained! [Video]. YouTube. https://www.youtube.com/watch?v=v6VJ2RO66A
-
Gutta, S. (2022, January 6). Folder Structure for Machine Learning Projects | by Surya Gutta | Analytics Vidhya. Medium. https://medium.com/analytics-vidhya/folder-structure-for-machine-learning-projects-a7e451a8caaa