• This repository represents " MultiVariate Classification ".
• With the help of this project we can Classifiy 4 Attributes of An Image .

📝 Description

• This implemantation is based on official resnet50
• In this project we have used Pretrained Model and tensorboard for image classification and checking the accuracy of the model.

⏳ Dataset

• Download the dataset for custom training
• https://drive.google.com/file/d/1mV7EP-maKTNu2RNv6wYRnaoON9dOqLNt/view?usp=sharing

🖥️ Installation

🛠️ Requirements

• Python 3.8+
• Tensorflow 2.9.1
• Keras
• Pandas
• Numpy
• Os

⚙️ Setup

1. Create virtual enviroment

$ conda create --prefix ./env python=3.8.13 -y

2. Activate conda enviroment

$ conda activate ./env

3. Install Required libraries

$ pip install requirements.txt

4. Run setup.py

$ pip install -e.

5. Run src/infrence.py To get the prediction.

🎯 Inference demo

1. Testing with Images (Put test inages in anywhere and give the location of this image to img_path parameter inside prediction model function in src/infrence.py file)

$ python src/infrence.py 

In img_path give the path of the image that you want to get prediction.

To run Tensorboard

tensorboard --logdir ./logs

Data Augumentation

For Data Augumentation We can use

• changes in Angels, Rotation and lighting
• Changes in Lighting and direction + Flipping about the vertical axis
• Flipping about the horizontal axis and rotating by 90, 180, 270 degress.

Contributor

• Sanjeev Kumar