Ai Package Wrapper

This is a wrapper package for multiple ai tools

It contains 4 total modules:

1. OCR Module
2. Augmentation Module
3. Sentiment Analysis Module
4. Explainable AI Module

OCR MODULE (made by @Madhavseekri)

This is a simple and efficient OCR (Optical Character Recognition) module written in Python. It allows you to extract text from images using pytesseract and Pillow.

What It Does

• Extracts readable text from image files.
• Supports PNG, JPEG, and other image formats.
• Can be extended for use in document scanning, data entry automation, and AI projects.

How It Works

1. extract_text(image_path)

Opens an image file and extracts text from it using pytesseract. Returns the extracted text.

Augmentation Module (made by @Ayushi-000)

augmentation is a Python package built for learning how to deploy real-world Python projects using Pip and Poetry. It includes simple examples of data augmentation for image, text, and audio files.

This package was created and deployed as part of a Prodigal's first task on Python package deployment using *Poetry, **PyPI, **GitHub, and *CI/CD pipelines.

How It Works

1. init()

Sets up three augmentation pipelines:

Audio Augmentation: Applies Gaussian noise, time stretching, and pitch shifting.

Image Augmentation: Uses horizontal flipping, random brightness/contrast, and rotation.

Text Augmentation: Performs synonym replacement using nlpaug.

2. augment_audio(input_path, output_path)

Reads an audio file, applies the audio augmentation pipeline, writes the augmented audio to a file, and returns the output path.

3. augment_image(input_path, output_path)

Loads an image, applies image augmentation, writes the augmented image back to disk, and returns the output path.

4. augment_text(text)

Processes a given text string through the text augmentation pipeline and returns the augmented text (a list of augmented strings).

Sentiment Analyzer (made by @abhay-cerberus)

A simple sentiment analysis package that leverages TextBlob to determine whether a piece of text expresses positive, negative, or neutral sentiment. It provides an easy-to-use interface for analyzing both single texts and batches of texts.

How It Works

1. analyze_text(text)

Processes a single text string by converting it to lowercase, calculating its sentiment polarity using TextBlob, and returning a sentiment label ("positive", "negative", or "neutral").

2. analyze_batch(texts)

Takes a list of text strings, analyzes each one using analyze_text, and then returns both the raw counts and percentages for each sentiment category.

ExplainableAi (made by @priyanshibindal)

How It Works

1. explain_model(model, data, method='shap')

Loads a pre-saved model and dataset (from churn_model.pkl), creates a SHAP explainer for the model, and generates visual explanations:

Produces a waterfall plot for the first prediction.

Produces a beeswarm plot summarizing feature importance.