Project 2

Transfer Learning

Cassava Disease Detection

Overview

Fine-tune pretrained models for cassava leaf disease detection. This project demonstrates transfer learning techniques using models like EfficientNet and ResNet, with callbacks for early stopping and model checkpointing.

Fine-tuning Pretrained Models Callbacks Images

Notebooks

Notebook Topic Description
023 Multiclass Classification Disease category prediction across multiple classes
024 Transfer Learning Fine-tuning pretrained models for domain adaptation
025 Callbacks Early stopping, model checkpointing, learning rate scheduling

Available Models

Model File Description
Pretrained Base pretrained_model.pth Base pretrained weights (ImageNet)
Fine-tuned model_trained.pth Fine-tuned on cassava dataset

Supported Architectures

EfficientNetB0 EfficientNetB3 ResNet50 VGG16

Available Tools

Tool Description
load_proj2_info Get Project 2 information and available resources
list_proj2_models List available pre-trained models for transfer learning