This is for the 2025Fall COMP4432 Machine Learning Course Project @ PolyU
Jupyter Notebook
3
6 commits
updated Apr 22, 2025
 |
|
PolyU 2025 Spring COMP4432 Team 24 Course Project
LogoCleaner is an AI-powered tool for detecting and removing logos from images. This repository contains multiple implementations with increasing sophistication, from traditional computer vision methods to state-of-the-art deep learning approaches. The project also includes a complete web application with an intuitive user interface for logo removal.
| Name | Student ID | Responsibility | Contribution |
|---|---|---|---|
| DAI Yuhang | 22097845D | Project Management, Model Training v2 | 20% |
| HE Yiyang | 22100143D | Model Training & Implementation v0, Backend API & Integration | 20% |
| YU Pengfei | 22102031D | Report Management & Model Implementation v1 | 20% |
| YAO Yuhao | 22097371D | Slides Design, Illustration Build & Model Implementation v2 | 20% |
| CHEN Fengyuan | 23096069D | Model Training v1 & Frontend Design | 20% |
Our project tackles the practical problem of unwanted logos in images (copyright issues, visual distraction) with a novel approach:
The LogoCleaner system provides several benefits:
Our web application makes advanced AI technology accessible to users without technical knowledge. The demo showcases real-world examples of logo removal from various images.
This project demonstrates several advanced technical skills:
The codebase is well-structured, documented, and follows best practices for each technology used.
We thoroughly evaluated each implementation:
The demo folder contains a complete web application for LogoCleaner:
Upload Image: Drag and drop or select an image containing a logo
Detection Phase: The system automatically detects logos in the image
Removal Process: Watch as the logo is removed and background is inpainted
Final Result: Download or share the logo-free image
For detailed instructions, see:
For the demo application, download the model from:
The selector module was trained on FlickrLogos-32 dataset released in ICMR11 and updated in ICML2017, which contains photos showing brand logos and is meant for the evaluation of logo retrieval and multi-class logo detection/recognition systems on real-world images.
Note that even though the dataset is open-source and well-known, we cannot provide the dataset with link since the owner requires an (informal) email to email in order to get the dataset. Our team also send the email and get the original datasets and we apologize for the inconvience.
Our baseline approach uses traditional computer vision techniques:
Read more about the SuperPixel + RF approach Setup
Our second approach leverages deep learning for more accurate segmentation:
Read more about the U-Net approach Setup
Our most advanced approach combines state-of-the-art models:
Read more about the SAM + NN approach Setup
We use a single RTX 4090 graphic card and train with 50 epoches (the selector module converges very fast at around the 45 epoch). You could refer to the training source code with our GitHub Repository.
We choose the classic fusion loss (BCE and Dice loss) for the evaluation results and outperforms the classic Unet with 2% dice loss. The details please refer to the report in our GitHub Repository.
from huggingface_hub import hf_hub_download
# Download the SAM model
sam_path = hf_hub_download(
repo_id="PeterDAI/LogoCleaner",
filename="sam_vit_b_01ec64.pth"
)
# Download the selector model
selector_path = hf_hub_download(
repo_id="PeterDAI/LogoCleaner",
filename="best_model.pth"
)
git clone https://github.com/hiteacherIamhumble/LogoCleaner.git
cd LogoCleaner
conda create -n logocleaner python=3.10
conda activate logocleaner
Follow each subfolder README.md for details
Datasets
cd demo/backend
pip install -r requirements.txt
python app.py
cd demo/frontend
npm install
npm run dev
Open your browser and navigate to http://localhost:8080
Upload an image, adjust settings if needed, and click "Remove Logo"
We welcome contributions to LogoCleaner! Please follow these steps:
git checkout -b feature/amazing-feature)git commit -m 'Add some amazing feature')git push origin feature/amazing-feature)For major changes, please open an issue first to discuss what you would like to change.
This project is licensed under the MIT License - see the LICENSE file for details.
This is for the 2025Fall COMP4432 Machine Learning Course Project @ PolyU
Jupyter Notebook
3
6 commits
updated Apr 22, 2025
|
|
PolyU 2025 Spring COMP4432 Team 24 Course Project
LogoCleaner is an AI-powered tool for detecting and removing logos from images. This repository contains multiple implementations with increasing sophistication, from traditional computer vision methods to state-of-the-art deep learning approaches. The project also includes a complete web application with an intuitive user interface for logo removal.
| Name | Student ID | Responsibility | Contribution |
|---|---|---|---|
| DAI Yuhang | 22097845D | Project Management, Model Training v2 | 20% |
| HE Yiyang | 22100143D | Model Training & Implementation v0, Backend API & Integration | 20% |
| YU Pengfei | 22102031D | Report Management & Model Implementation v1 | 20% |
| YAO Yuhao | 22097371D | Slides Design, Illustration Build & Model Implementation v2 | 20% |
| CHEN Fengyuan | 23096069D | Model Training v1 & Frontend Design | 20% |
Our project tackles the practical problem of unwanted logos in images (copyright issues, visual distraction) with a novel approach:
The LogoCleaner system provides several benefits:
Our web application makes advanced AI technology accessible to users without technical knowledge. The demo showcases real-world examples of logo removal from various images.
This project demonstrates several advanced technical skills:
The codebase is well-structured, documented, and follows best practices for each technology used.
We thoroughly evaluated each implementation:
The demo folder contains a complete web application for LogoCleaner:
Upload Image: Drag and drop or select an image containing a logo
Detection Phase: The system automatically detects logos in the image
Removal Process: Watch as the logo is removed and background is inpainted
Final Result: Download or share the logo-free image
For detailed instructions, see:
For the demo application, download the model from:
The selector module was trained on FlickrLogos-32 dataset released in ICMR11 and updated in ICML2017, which contains photos showing brand logos and is meant for the evaluation of logo retrieval and multi-class logo detection/recognition systems on real-world images.
Note that even though the dataset is open-source and well-known, we cannot provide the dataset with link since the owner requires an (informal) email to email in order to get the dataset. Our team also send the email and get the original datasets and we apologize for the inconvience.
Our baseline approach uses traditional computer vision techniques:
Read more about the SuperPixel + RF approach Setup
Our second approach leverages deep learning for more accurate segmentation:
Read more about the U-Net approach Setup
Our most advanced approach combines state-of-the-art models:
Read more about the SAM + NN approach Setup
We use a single RTX 4090 graphic card and train with 50 epoches (the selector module converges very fast at around the 45 epoch). You could refer to the training source code with our GitHub Repository.
We choose the classic fusion loss (BCE and Dice loss) for the evaluation results and outperforms the classic Unet with 2% dice loss. The details please refer to the report in our GitHub Repository.
from huggingface_hub import hf_hub_download
# Download the SAM model
sam_path = hf_hub_download(
repo_id="PeterDAI/LogoCleaner",
filename="sam_vit_b_01ec64.pth"
)
# Download the selector model
selector_path = hf_hub_download(
repo_id="PeterDAI/LogoCleaner",
filename="best_model.pth"
)
git clone https://github.com/hiteacherIamhumble/LogoCleaner.git
cd LogoCleaner
conda create -n logocleaner python=3.10
conda activate logocleaner
Follow each subfolder README.md for details
Datasets
cd demo/backend
pip install -r requirements.txt
python app.py
cd demo/frontend
npm install
npm run dev
Open your browser and navigate to http://localhost:8080
Upload an image, adjust settings if needed, and click "Remove Logo"
We welcome contributions to LogoCleaner! Please follow these steps:
git checkout -b feature/amazing-feature)git commit -m 'Add some amazing feature')git push origin feature/amazing-feature)For major changes, please open an issue first to discuss what you would like to change.
This project is licensed under the MIT License - see the LICENSE file for details.