CropLens - AI

Flower Classification & Dataset Pipeline (3-Class)

An end-to-end Computer Vision project for flower classification and dataset engineering, designed for research and real-world AI applications.

Overview

This project focuses on building a robust flower dataset (3 classes) and applying deep learning techniques for classification and preprocessing.

It includes:

1.Dataset collection & cleaning;

2.Background removal pipeline;

3.Data augmentation;

4.Model training & evaluation.

Objectives:

1.Build a high-quality flower dataset;

2.Remove noisy backgrounds for better feature learning;

3.Train a high-performance classification model;

4.Provide a reproducible pipeline for research use.

Preview

Methodology:

1. Data Preprocessing

Background removal using OpenCV / rembg; Image resizing (e.g., 224x224); Noise filtering.

2. Data Augmentation

Rotation; Flipping; Brightness adjustment.

3. Model Architecture

CNN / Transfer Learning; Loss: CrossEntropyLoss; Optimizer: Adam.