Update README.md
Browse files
README.md
CHANGED
|
@@ -1,2 +1,78 @@
|
|
| 1 |
-
|
| 2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
tags:
|
| 4 |
+
- source-free
|
| 5 |
+
- object-detection
|
| 6 |
+
- domain-adaptation
|
| 7 |
+
- medical-imaging
|
| 8 |
+
- semi-supervised
|
| 9 |
+
- computer-vision
|
| 10 |
+
model-index:
|
| 11 |
+
- name: Grounded Teacher
|
| 12 |
+
results: []
|
| 13 |
+
paper:
|
| 14 |
+
- name: Context Aware Grounded Teacher for Source Free Object Detection
|
| 15 |
+
url: https://github.com/Tajamul21/Grounded_Teacher
|
| 16 |
+
---
|
| 17 |
+
|
| 18 |
+
# Grounded Teacher (GT)
|
| 19 |
+
|
| 20 |
+
This repository contains model weights and resources for **Grounded Teacher**, a framework designed to tackle **Source Free Object Detection (SFOD)**, especially under **domain shift** and **context imbalance** conditions—common challenges in **medical imaging**.
|
| 21 |
+
|
| 22 |
+
---
|
| 23 |
+
|
| 24 |
+
## 🧠 Abstract
|
| 25 |
+
|
| 26 |
+
We focus on the Source Free Object Detection (SFOD) problem, where **source data is unavailable** during adaptation, and the model must adapt to an **unlabeled target domain**. In medical imaging, many approaches rely on semi-supervised student-teacher setups to handle domain gaps.
|
| 27 |
+
|
| 28 |
+
However, **context imbalance** and **domain shifts** often lead to **biased teacher models**, producing poor pseudolabels and causing the student model to collapse.
|
| 29 |
+
|
| 30 |
+
To address this, we introduce **Grounded Teacher (GT)**:
|
| 31 |
+
- Models **contextual relationships** via a **relational context module**.
|
| 32 |
+
- Applies targeted **augmentations** for underrepresented classes across/within domains.
|
| 33 |
+
- Adds an **expert foundational branch** to supervise the student model and enhance prediction quality.
|
| 34 |
+
|
| 35 |
+
GT shows significant improvement across three medical imaging datasets, with thorough ablation studies validating its effectiveness.
|
| 36 |
+
|
| 37 |
+
---
|
| 38 |
+
|
| 39 |
+
## 📦 Resources
|
| 40 |
+
|
| 41 |
+
- 🔗 **GitHub (code, data, training scripts):** [Grounded Teacher on GitHub](https://github.com/Tajamul21/Grounded_Teacher)
|
| 42 |
+
|
| 43 |
+
---
|
| 44 |
+
|
| 45 |
+
## 👨🔬 Authors
|
| 46 |
+
|
| 47 |
+
- **Tajamul Ashraf**
|
| 48 |
+
- Rajes Manna
|
| 49 |
+
- Partha Sarathi Purkayastha
|
| 50 |
+
- Tavaheed Tariq
|
| 51 |
+
- Janibul Bashir
|
| 52 |
+
|
| 53 |
+
---
|
| 54 |
+
|
| 55 |
+
## 📅 Publication
|
| 56 |
+
|
| 57 |
+
**Title:** Context Aware Grounded Teacher for Source Free Object Detection
|
| 58 |
+
**Published:** April 22
|
| 59 |
+
|
| 60 |
+
---
|
| 61 |
+
|
| 62 |
+
## 📃 Citation
|
| 63 |
+
|
| 64 |
+
```bibtex
|
| 65 |
+
@inproceedings{groundedteacher2025,
|
| 66 |
+
title={Context Aware Grounded Teacher for Source Free Object Detection},
|
| 67 |
+
author={},
|
| 68 |
+
booktitle={},
|
| 69 |
+
year={2025}
|
| 70 |
+
}
|
| 71 |
+
```
|
| 72 |
+
|
| 73 |
+
---
|
| 74 |
+
|
| 75 |
+
## 🔍 License
|
| 76 |
+
|
| 77 |
+
MIT License
|
| 78 |
+
|