Image Recognition Model Task

This project is a Deep Learning task designed to classify images into two main categories: Linear and Circular.

🛣️ Project Workflow

The project was developed through the following steps:

1. Data Collection & Training: The pipeline started on Teachable Machine, where a custom dataset was gathered and structured into two classes: Linear and Circular. A collection of sample images was uploaded for each class to train the vision model.
2. Model Export: Once trained, the trained weights were exported as a Keras model (keras_model.h5) along with its respective labels.txt.
3. Inference & Testing: A Python script was developed using tf_keras and Pillow to load the trained model, preprocess new test images, and validate the prediction accuracy locally.

🚀 Features

• Binary Classification: Accurately distinguishes between Linear and Circular shapes in images.
• Data-Driven: Trained on a dedicated dataset compiled specifically for this task.
• Confidence Scoring: Displays the predicted class along with the model's prediction confidence score.

📋 Requirements

Before running the script, make sure you install the required dependencies: pip install tensorflow tf_keras pillow numpy

🛠️ How to Use

Clone or download this repository.

Make sure keras_model.h5, labels.txt, and your test image (test_image.png) are in the same folder.

Run the prediction script.

📊 Sample Output

When feeding a circular shape image, the expected output structure is:

Class: circular

Confidence Score: 0.9967167