FrereAlidor/LSC-Lingala-sign-Efficient-LSTM

# LSC-Lingala-sign-Efficient-LSTM# Intelligent Lingala Sign Language Translation System **Developed for the Democratic Republic of Congo** A state-of-the-art computer vision system that translates Lingala sign language gestures into text and speech using deep learning technologies. This system combines EfficientNet-B3 architecture with LSTM networks for temporal gesture modeling, providing real-time translation capabilities. ## 🌟 Features - **Real-time Translation**: Live webcam-based sign language recognition - **Advanced Architecture**: EfficientNet-B3 + LSTM for superior accuracy - **MediaPipe Integration**: Enhanced hand landmark detection and preprocessing - **Multi-modal Output**: Text and speech synthesis capabilities - **Robust Evaluation**: K-fold cross-validation and comprehensive metrics - **User-friendly Interface**: Interactive demonstration system - **Data Augmentation**: Advanced preprocessing for improved model generalization ## 🏗️ Architecture ### Model Components 1. **Feature Extraction**: EfficientNet-B3 (pre-trained on ImageNet) 2. **Temporal Modeling**: LSTM layers for sequence understanding 3. **Hand Detection**: MediaPipe for precise hand landmark identification 4. **Classification**: Dense layers with softmax activation ### Architecture Variants - **Single Image Model**: Direct EfficientNet-B3 classification - **Sequence Model**: TimeDistributed EfficientNet + LSTM for temporal sequences ## 🚀 Getting Started ### Prerequisites ```bash pip install tensorflow>=2.8.0 pip install opencv-python pip install mediapipe pip install scikit-learn pip install matplotlib pip install seaborn pip install pandas pip install torch pip install gtts ``` ### Installation 1. Clone the repository: ```bash git clone github.com cd lingala-sign-language-translation ``` 2. Prepare your data structure: ``` Data_TeachSign/ ├── X_train.npy ├── y_train.npy ├── X_val.npy ├── y_val.npy └── labels.npy …