lucasxlu / FGVC

Fine-Grained Visual Classification for Plants & Flowers

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Fine-Grained Visual Classification

Introduction

  1. Train ResNet to recognize over 998 categories (997 plants + 1 others) for Kaggle Competition.

  2. Deep Learning for Plants Disease Recognition for AI Challenger 2018 Competition.

Solutions

  • Directly train state-of-the-art models (ResNeXt/ResNet/DenseNet) pretrained on ImageNet and fine-tune for classification.
  • Apply bilinear pooling and zoom data augmentation for fine-grained visual recognition.

Results

Model Top-1 Accuracy
ResNet50 89.0947%
ResNet18 84.5452%

About

Fine-Grained Visual Classification for Plants & Flowers

License:MIT License


Languages

Language:Python 100.0%