LingualSense_Infosys_Internship_Oct2024

To build a model that can automatically identify the language of a given text. Language identification is essential for various applications, including machine translation, multilingual document tracking, and electronic devices (e.g., mobiles, laptops).

LingualSense: Deep Learning for Language Detection Across Texts

LingualSense is a deep learning project for classifying text languages. This README provides step-by-step instructions from data analysis to deployment.

---

Steps to Follow

1. Exploratory Data Analysis (EDA)

• Perform EDA to analyze your dataset.
• Check the distribution of languages and clean any irregularities in the dataset.

2. Data Preprocessing

• Tokenize the text data and pad sequences to a uniform length for model compatibility.
• Save the tokenizer and label encoder for future use in the app.

3. Model Building

• Use a GRU-based model for text classification.
• Train the model using tokenized and padded sequences.
• Save the trained model as gru_model.h5.

4. Streamlit Application Development

• Create a Streamlit app for real-time predictions.
• Include input text areas, model loading, and prediction functionality.
• Add a styled user interface for better interaction.

5. Setup Environment

1. Clone the repository:

   git clone https://github.com/Springboard429/LingualSense_Infosys_Internship_Oct2024.git
   cd LingualSense
   

2. Create a virtual environment:

   • Windows:

     python -m venv lingualsense_env
     lingualsense_env\Scripts\activate

   • Mac/Linux:

     python -m venv lingualsense_env
     source lingualsense_env/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. Place the following files in the project directory:

    gru_model.h5
    tokenizer.joblib
    label_encoder.joblib
   

6. Run the Streamlit App

• Execute the following command:

  streamlit run app.py

Open the local URL (e.g., http://localhost:XXXX) to access the app.

7. Usage

• Input text in the text area.
• Click "Detect Languages" to get the predicted language of the text.