[Python + Tensorflow] Pests Classification

Introduction

Using Transfer Learning with VGG16, Resnet50, Inception v3 and Flask Framework to build a website that distinguishes 9 types of insects : armyworm, beetle, cicadellidae, cricket, grasshopper, limacodidae, lycorma delicatula, mosquito, weevil

How to use my project

Step 1 :

Clone my project

git clone https://github.com/thangnh1/PestClassification

Step 2 :

Open with editor tool, install lib with terminal

pip install -r requirements.txt

Step 3 :

Train model (Recommend using Google Colab)

Upload folder PestClassification/data to Google Drive, then Open 3 file from PestClassification/notebook/ with Google Colab, connect GPU and Run All

Step 4 :

Copy le.pkl to PestClassification and create folder models, copy 2 file .hdf5 from Google Drive to PestClassification/models/ . You can test models with predict.py file or evaluate models with score.py

Step 5 :

Open Terminal, run command python server.py

Demo

Data Sample

Result Evaluate

Main Screen

Choose Model

Choose Image

Result