Live Virtual AI Trainer

Job ID: 38969984

Budget: ₹600 – ₹3,000 INR

**Project Title:** Virtual AI Trainer Using MediaPipe (Live Video Pose Detection)

**Project Description:**
I am looking for a freelancer to develop a **Virtual AI Trainer** that uses the **MediaPipe Pose** model (pre-trained) for real-time video-based exercise detection and counting. The project focuses on standing exercises, primarily using dumbbells, and aims to provide accurate feedback based on user posture and movements.

### Key Requirements:
1. **Pose Detection with MediaPipe**:
- Utilize the pre-trained **MediaPipe Pose** model for real-time video pose detection.
- Accurately detect keypoints and angles to evaluate exercises.

2. **Live Video Input**:
- The system should work with a live webcam feed for real-time exercise monitoring.

3. **Exercises**:
Implement detection and counting for **6 simple standing exercises**, including but not limited to:
- Dumbbell Bicep Curls
- Dumbbell Shoulder Press
- Dumbbell Side Lateral Raise
- Squats
- Lunges
- Standing Dumbbell Row or any standing exercises to show for the project review.

4. **Feedback Mechanism**:
- Provide visual or textual feedback on posture (e.g., “Straighten your back” or “Lower your arms more”).
- Count repetitions automatically.

5. **Simple UI**:
- A minimalistic interface to display:
- Live video feed.
- Exercise instructions.
- Rep counts.

6. **Customization**:
- Easy addition of new exercises or adjustment of existing thresholds (e.g., angle ranges).

### Deliverables:
1. Complete source code with clear documentation.
2. Instructions on how to set up and run the project (preferably compatible with PyCharm).
3. A brief testing report to verify the accuracy of each exercise.

### Additional Notes:
- The project is **simple**, leveraging **pre-trained MediaPipe models** to minimize complexity.
- Accurate detection for standing exercises is a priority, as lying or seated exercises won't be included.
- If the freelancer can suggest improvements to enhance functionality or accuracy, they are welcome.

---

Budget: under 3000
Timeline: I am a pg student, Every monday I have project presentation. So need to showcase something on every monday.
Preferred Skills:
- Python programming.
- Experience with **MediaPipe** or similar pose estimation frameworks.
- Knowledge of real-time video processing (OpenCV or similar).
Related categories: Python UI / User Interface