kkthyagharajan's picture
Upload README.md
f23ef33 verified
---
language: en
library_name: keras
tags:
- image-classification
- transfer-learning
- computer-vision
- keras
- tensorflow
- multiclass-classification
license: mit
datasets:
- custom
model-type: multi-model-repository
author: Thyagharajan K K
pipeline_tag: image-classification
inference: true
app_file: app.py
---
# 🧠 KKT-HF-TransferLearning-Models
## **Ready-to-Use Transfer Learning Models for Image Classification**
**Created by [Thyagharajan K K](https://huggingface.co/kkthyagharajan)**
[![Hugging Face Hub](https://img.shields.io/badge/HuggingFace-Repository-orange?logo=huggingface)](https://huggingface.co/kkthyagharajan/KKT-HF-TransferLearning-Models)
[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
[![Made with Keras](https://img.shields.io/badge/Made%20with-Keras-red?logo=keras)](https://keras.io/)
[![TensorFlow](https://img.shields.io/badge/Framework-TensorFlow-orange?logo=tensorflow)](https://www.tensorflow.org/)
[![KKT_DL_Package](https://img.shields.io/badge/Package-KKT__DL__Package-blue?logo=python)](https://github.com/kkthyagharajan/KKT_DL_Package)
---
## πŸ“˜ Overview
This repository hosts a **collection of pretrained image-classification models** created using **Transfer Learning** in Keras/TensorFlow.
Each subdirectory contains:
- A trained `.keras` model file
- A `class_names.txt` file
- A `Test` folder with example test images
You can use these models **programmatically** or through an **interactive demo app** powered by Gradio or Streamlit.
---
## 🧩 Programmatic Usage
```python
from KKT_DL_Package.utils.KKT_predictions import (
multiclass_prediction_return,
display_images_gui,
get_hf_model_img_labels_local_path,
)
IMG_SIZE = (300, 300)
model_full_path, test_folder_path, class_names = (
get_hf_model_img_labels_local_path(
"kkthyagharajan/KKT-HF-TransferLearning-Models",
IMG_SIZE,
force_refresh=False, # Won’t redownload if cached
)
)
# Run predictions and display results
all_image_paths, all_predicted_labels, all_confidences = (
multiclass_prediction_return(
model_full_path,
test_folder_path,
class_names,
IMG_SIZE,
)
)
display_images_gui(
all_image_paths,
all_predicted_labels,
IMG_SIZE,
)
```
---
## πŸš€ Interactive Demo App
### 🧩 Option 1: Run directly on Hugging Face
This Space includes a web app defined by `app.py`.
### πŸ’» Option 2: Run locally using Gradio or Streamlit
```bash
pip install -r requirements.txt
python app.py # or
streamlit run app.py
```
---
## πŸ“ Repository Structure
```
KKT-HF-TransferLearning-Models/ ← Root directory (your HF repo root)
β”‚
β”œβ”€β”€ Insect_Inception_V3/ ← Model 1 directory
β”‚ β”œβ”€β”€ insect_inception_v3_model.keras ← Saved model file
β”‚ β”œβ”€β”€ class_names.txt ← Corresponding class labels
β”‚ └── InsectTest/ ← Test image folder
β”‚ β”œβ”€β”€ image_001.jpg
β”‚ β”œβ”€β”€ image_002.jpg
β”‚ └── ...
β”‚
β”œβ”€β”€ Insect_MobileNetV2/ ← Model 2 directory (example)
β”‚ β”œβ”€β”€ insect_mobilenet_v2_model.keras
β”‚ β”œβ”€β”€ class_names.txt
β”‚ └── InsectTest/
β”‚ β”œβ”€β”€ test1.jpg
β”‚ β”œβ”€β”€ test2.jpg
β”‚ └── ...
β”‚
β”œβ”€β”€ Insect_ResNet50/
β”‚ β”œβ”€β”€ insect_resnet50_model.keras
β”‚ β”œβ”€β”€ class_names.txt
β”‚ └── InsectTest/
β”‚ β”œβ”€β”€ imgA.jpg
β”‚ β”œβ”€β”€ imgB.jpg
β”‚ └── ...
β”‚
└── README.md
```