Multivariate Classification

📝 Description

This implementation 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.

STEPS -

STEP 01- Create a conda environment after opening the repository in VSCODE

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

STEP 02- install the requirements

pip install -r requirements.txt

STEP 03- initialize the dvc project

dvc init

It will run for 100 Epochs. You can change this into params.yaml section for the number of records.

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

⏳ Dataset

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

:desktop_computer: Installation

:hammer_and_wrench: Requirements

Python 3.8+
Tensorflow 2.9.1
Keras
Pandas
Numpy
Os

:gear: Setup

Create virtual environment.

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

Activate virtual enviroment.

conda activate ./env

OR

$ source activate ./env

Run setup.py
```
$ pip install -e.
```
Initialize DVC
```
$ dvc init
```
Run All the steps for DVC
```
$ dvc repro
```
After that our model will trained on the given dataset for 10 epochs. We can modify the parameters in the params.yaml file in the root directory of the folder.

🎯 Inference demo

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)

infrence_example

$ 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 Augmentation

For Data Augmentation 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