Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/anidipta/quarter-ball-tracker-using-yolo
Ball Detection in quarter
https://github.com/anidipta/quarter-ball-tracker-using-yolo
colab-notebook computer-vision opencv pillow python
Last synced: 7 days ago
JSON representation
Ball Detection in quarter
- Host: GitHub
- URL: https://github.com/anidipta/quarter-ball-tracker-using-yolo
- Owner: Anidipta
- License: apache-2.0
- Created: 2024-07-07T17:59:02.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2024-08-19T16:27:40.000Z (5 months ago)
- Last Synced: 2024-08-19T19:29:29.119Z (5 months ago)
- Topics: colab-notebook, computer-vision, opencv, pillow, python
- Homepage:
- Size: 1.53 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Project Title: Ball ⚽ Tracking and Event Detection in Video 🎥
## **Objective:**
Develop an advanced computer vision program to track and classify the movement of balls in various colors across different quadrants within a video. The system will record each ball's entry and exit events from each numbered quadrant, displaying this information overlayed on the processed video.
## **Model Details:**
Two custom YOLOv8x models were employed in this project:
- **🔲 Quadrant Detection Model**: Identifies quadrants in the video to accurately map ball positions.
- **🏀 Ball Detection Model**: Detects balls, classifies their colors, and tracks their movements to log entry and exit events.
- **⚙️ Combined Model**: Integrates the above models using a cascading object detection approach for comprehensive detection and identification.
## **Links and Results:**
- **📔 Link to the Colab Notebook**: [CLICK HERE](https://colab.research.google.com/drive/1EGUlv8hCES5XOHuVv5tpenP1wSIR0ry1?usp=sharing)
### **🔍 Model Results:**
- **Confusion Matrices**
- **Ball Detection Model**:
📊 Ball Detection Model
.png)
📊 Quadrant Detection Model
- **Label Analysis**
- **Ball Detection Model**:
📈 Ball Detection Model
.jpg)
📈 Quadrant Detection Model
- **Detection Results**
- **Ball Detection Model**:
🔍 Ball Detection Model
.png)
🔍 Quadrant Detection Model
## **Output:**
### **🎬 Processed Video:**
- Tracks balls with color identification.
- Displays time stamps in the top left corner.
- Shows detection boxes with confidence levels.
- **🔗 Link to the Processed Video**: [CLICK HERE](https://drive.google.com/file/d/1c_EzHK5AWmWOBoT0Yf4Q2Zuz8NICl0Jl/view?usp=sharing)
- **🖼️ Demo Image Detected by the Model:**
### **📄 Text File:**
- Records events with format: Time, Quadrant Number, Ball Color, Type (Entry or Exit).
- **🔗 Link to the TXT File**: [CLICK HERE](cleaned_result.txt)
## **Tools and Technologies Used:**
- **🧠 YOLOv8x** for advanced object detection.
- **🔧 OpenCV** for video processing.
- **🔢 NumPy** for efficient data manipulation.
- **📊 Pandas** for data management and analysis.
- **🐍 Python** for scripting and integration.
- **🔒 Kaggle** for storing videos and models.
- **🗂️ Roboflow** for dataset preparation (training, testing, and validation).
## **Challenges Faced:**
1. **🔄 Object Tracking and Identification**: Maintaining ball identity across frames despite occlusions and rapid movement.
2. **⏱️ Real-time Performance**: Processing each video frame in real-time while managing complex models.
3. **🎯 Accuracy of Detection**: Achieving high detection accuracy under varied lighting and backgrounds.
4. **🔄 State Management**: Correctly updating the state of balls while tracking multiple objects in dynamic scenes.
## **Usage Instructions:**
1. Clone the repository and navigate to the project directory.
2. Ensure all required packages are installed by running:
```bash
pip install -r requirements.txt
```
3. Place your input video in the project directory.
4. Update the paths in the `process_video` function to point to your input video and desired output locations.
5. Run the Jupyter notebook to process the video and generate output files.