DCNN Plant Classification Project

Overview

This project involved implementing and improving a Deep Convolutional Neural Network (DCNN)-based plant image classification model using TensorFlow. The objective was to classify 12 different plant species by performing data preprocessing, designing various DCNN architectures, and evaluating their performance. Additionally, the impact of model depth on classification accuracy was analysed.

Results

Baseline DCNN model: Test accuracy 68.2%

Improved DCNN model: Test accuracy 69.7%

Optimised DCNN model: Test accuracy 70.3%

Tech Stack

Python, DCNN model design and implementation, data processing and preprocessing, model training and optimisation, performance evaluation and analysis

Code

The full code can be accessed below.

GitHub — source

Report

You can check the project report here — Korean report available below.

Download