Update README.md

README.md

@@ -1,9 +1,12 @@
+---
+datasets:
+- xiaoqi-wang/miqd-2.5m
+---
 
 <div align="center">
   <img src="./assets/logo1.png" alt="IQA Logo" width="1200"/>
-  
-  <h3><strong> Image Quality Assessment for Machines: Paradigm, Large-scale Database, and Models
-</strong></h3> 
+  <h2><strong> Image Quality Assessment for Machines: Paradigm, Large-scale Database, and Models
+</strong></h2> 
 
   [![Database](https://img.shields.io/badge/Database-Available-green?style=flat-square)](https://huggingface.co/datasets/xiaoqi-wang/miqd-2.5m)
   [![Paper](https://img.shields.io/badge/arXiv-Paper-red?style=flat-square)](https://arxiv.org/abs/2508.19850)
@@ -11,32 +14,31 @@
   [![GitHub Stars](https://img.shields.io/github/stars/XiaoqiWang/MIQA?style=social)](https://github.com/XiaoqiWang/MIQA)
   
 [中文](README_CN.md) | [English](README.md) | [Model](https://huggingface.co/xiaoqi-wang/miqa) | [Dataset](https://huggingface.co/datasets/xiaoqi-wang/miqd-2.5m)
+
 </div>
-<div style="font-size: 13px;">
+<div style="font-size: 15px;">
 🎯 Project Overview
-
+  
 - 🤖 Machine-Centric: We bypass human perception to evaluate images from the perspective of the deep learning models that use them.  
 - 📈 Task-Driven Metrics: Directly measure how degradations like blur, noise, or compression artifacts impact the performance of downstream vision tasks.  
 - 💡 A New Paradigm: MIQA offers a new lens for optimizing image processing pipelines where machines make the final decision.
 </div>
 
----
 ## ✨ Does MIQA Work?
-<div align="center">
-  <img src="./assets/cls_ratio.png" alt="Classification Performance" width="30%"/>
-  <img src="./assets/det_ratio_ap75.png" alt="Detection Performance" width="30%"/>
-  <img src="./assets/ins_ratio_ap75.png" alt="Instance Segmentation Performance" width="30%"/>
-  <p><em>Performance improvement across tasks when filtering low-quality images using MIQA scores</em></p>
-</div>
-
+<table>
+<tr>
+<td><img src="./assets/cls_ratio.png" width="100%"></td>
+<td><img src="./assets/det_ratio_ap75.png" width="100%"></td>
+<td><img src="./assets/ins_ratio_ap75.png" width="100%"></td>
+</tr>
+</table>
+<p align="center"><em>Performance improvement across tasks when filtering low-quality images using MIQA scores</em></p>
 <details>
 <summary> 🗝️ Key Results</summary>
-
 Our results provide clear evidence of MIQA's effectiveness across three representative computer vision tasks: classification, detection, and segmentation.
 The framework consistently identifies images that degrade model performance. By filtering these detrimental samples, MIQA directly leads to improved outcomes and demonstrates the universal utility of a machine-centric approach. This transforms quality assessment from a passive metric into a proactive tool, safeguarding downstream models against the unpredictable image quality of real-world conditions and ensuring robust performance when it matters most.
 </details>
 
----
 ## 🛠️ Installation Guide
 
 #### Step 1: Install Dependencies
@@ -61,7 +63,6 @@ If your CUDA version is relatively high, such as 12.7 or higher, you might encou
 
 For example, if you need a specific version of **mmcv**, you can uninstall the existing versions and install a compatible one as follows:
 
-
 ```bash
 pip uninstall mmcv mmcv-full -y
 mim install "mmcv>=2.0.0rc4,<2.2.0"  # The version specified here is just an example. You should choose a version that is compatible with your CUDA and PyTorch setup.*
@@ -75,31 +76,40 @@ pip install -r requirements.txt
 ```
 
 ## 📦 Model Weights & Performance
-
 ### Where things live
+<table width="100%">
+<tr>
+<th align="left">Role</th>
+<th align="left">Location</th>
+</tr>
+
+<tr>
+<td><b>Application code</b> (training, inference, evaluation)</td>
+<td>This <b>GitHub</b> repository: <a href="https://github.com/XiaoqiWang/MIQA">github.com/XiaoqiWang/MIQA</a></td>
+</tr>
 
-| Role | Location |
-|------|----------|
-| **Application code** (training, inference, evaluation) | This **GitHub** repository: [github.com/XiaoqiWang/MIQA](https://github.com/XiaoqiWang/MIQA) |
-| **Published RA-MIQA checkpoints** (9 files) | **Hugging Face** model repo: [xiaoqi-wang/miqa](https://huggingface.co/xiaoqi-wang/miqa) |
-| **MIQD-2.5M database** | **Hugging Face** dataset: [xiaoqi-wang/miqd-2.5m](https://huggingface.co/datasets/xiaoqi-wang/miqd-2.5m) |
+<tr>
+<td><b>Published RA-MIQA checkpoints</b> (9 files)</td>
+<td><b>Hugging Face</b> model repo: <a href="https://huggingface.co/xiaoqi-wang/miqa">xiaoqi-wang/miqa</a></td>
+</tr>
 
+<tr>
+<td><b>MIQD-2.5M database</b></td>
+<td><b>Hugging Face</b> dataset: <a href="https://huggingface.co/datasets/xiaoqi-wang/miqd-2.5m">xiaoqi-wang/miqd-2.5m</a></td>
+</tr>
+</table>
 
 ### Naming & cache
 
-Checkpoint pattern on the Hub:
-
-`miqa_ra_miqa_{cls|det|ins}_{composite|consistency|accuracy}_metric.pth.tar`
+Checkpoint pattern on the Hub: `miqa_ra_miqa_{cls|det|ins}_{composite|consistency|accuracy}_metric.pth.tar`
 
 Examples:
-
 - `miqa_ra_miqa_cls_composite_metric.pth.tar`
 - `miqa_ra_miqa_det_consistency_metric.pth.tar`
 - `miqa_ra_miqa_ins_accuracy_metric.pth.tar`
 
 On first run, `huggingface_hub` downloads into `models/checkpoints/{composite|consistency|accuracy}_metric/`. 
 
-
 ## 🚀 Quick Start
 
 ### Assess a Single Image
@@ -132,7 +142,6 @@ python img_inference.py --input path/to/image.jpg --task cls --model ra_miqa --s
 python img_inference.py --input ./assets/demo_images/imagenet_demo --task ins --save-results --visualize
 ```
 
-
 ### 🎬 Video Assessment
 
 Video Quality Assessment offers two workflows: **(1) Frame-by-Frame Annotation**: Generates fully annotated videos for detailed visual inspection. Suitable for demos and qualitative analysis but computationally intensive.
@@ -159,9 +168,18 @@ The primary output is a new `.mp4` video file. This video shows the original foo
 
 <summary>🎥 <b>Example: Frame-wise MIQA Predictions on Videos</b></summary> 
 
-| Brightness Variation | Compression Artifacts | Minimal Perceptual Distortion |
-| :---: | :---: | :---: |
-| <video src="https://github.com/user-attachments/assets/9b20cbc4-3baf-4d57-8d5f-49acd6873725" width="280" controls></video> | <video src="https://github.com/user-attachments/assets/c2fc142b-6889-4451-8a05-fb93e0ec0656" width="280" controls></video> | <video src="https://github.com/user-attachments/assets/14f4fc37-5ae5-4068-81f2-6f86bec30a27" width="280" controls></video> | 
+<table>
+<tr>
+<td align="center"><b>Brightness Variation</b></td>
+<td align="center"><b>Compression Artifacts</b></td>
+<td align="center"><b>Minimal Perceptual Distortion</b></td>
+</tr>
+<tr>
+<td><video src="https://github.com/user-attachments/assets/9b20cbc4-3baf-4d57-8d5f-49acd6873725" width="100%" controls></video></td>
+<td><video src="https://github.com/user-attachments/assets/c2fc142b-6889-4451-8a05-fb93e0ec0656" width="100%" controls></video></td>
+<td><video src="https://github.com/user-attachments/assets/14f4fc37-5ae5-4068-81f2-6f86bec30a27" width="100%" controls></video></td>
+</tr>
+</table>
 
 </details>
 
@@ -182,7 +200,6 @@ This workflow is highly optimized for batch processing.
 # Analyze all videos in a directory, sampling 120 frames from each
 python video_analytics_inference.py --input assets/demo_video/ --task det --video-frames 120 --visualize
 
-
 python video_analytics_inference.py --input assets/demo_video/jpeg_distorted.mp4  --task det --visualize --viz-granularity both
 # viz-granularity both : Specifies the type of plot to generate. 'composite' creates a comprehensive, side-by-side comparison chart showing:
 #1. The raw, frame-level quality scores. 2. The smoothed, per-second average quality scores.
@@ -192,15 +209,22 @@ This process **does not create a new video**. Instead, it generates two key outp
 1.  A **`.png` image**: A detailed time-series plot showing the quality score fluctuation over the video's duration.
 2.  A **`.json` file**: A structured data file containing per-second aggregated scores, overall statistics (average, min, max, std. dev), and video metadata.
 
-
-
 <details>
 <summary>🎥 <b>Example: Frame-wise MIQA Predictions on Videos</b></summary>
 
-| Brightness Variation | Compression Artifacts |         Minimal Perceptual Distortion          |
-| :---: | :---: |:----------------------------------------------:|
-| <img src="inference_results/brightness_distorted_composite_quality_comparison.png" width="280"> | <img src="inference_results/jpeg_distorted_composite_quality_comparison.png" width="280"> | <img src="inference_results/B314_composite_quality_comparison.png" width="280"> |
- 
+<table>
+<tr>
+<td align="center"><b>Brightness Variation</b></td>
+<td align="center"><b>Compression Artifacts</b></td>
+<td align="center"><b>Minimal Perceptual Distortion</b></td>
+</tr>
+<tr>
+<td><img src="inference_results/brightness_distorted_composite_quality_comparison.png" width="100%"></td>
+<td><img src="inference_results/jpeg_distorted_composite_quality_comparison.png" width="100%"></td>
+<td><img src="inference_results/B314_composite_quality_comparison.png" width="100%"></td>
+</tr>
+</table>
+
 </details>
 
 ## 🏃 Training and Evaluation
@@ -762,4 +786,4 @@ If you find this work useful in your research, please consider citing:
 ## 📧 Contact
 
 - **Project Maintainer**: [Xiaoqi Wang](mailto:wangxq79@mail2.sysu.edu.cn)
-- **Issues**: Please use [GitHub Issues](https://github.com/XiaoqiWang/MIQA/issues) for bug reports and feature requests
+- **Issues**: Please use [GitHub Issues](https://github.com/XiaoqiWang/MIQA/issues) for bug reports and feature requests