Khubaib01/ConfiDetect-Confidence-Posture-Dataset

🧍‍♂️ ConfiDetect Confidence Posture Dataset

Subtitle

Pose-based Confidence Estimation Dataset for Behavioral and Affective Computing

---

📘 Overview

The ConfiDetect Confidence Posture Dataset contains normalized geometric and ratio-based features extracted from human posture and facial keypoints to classify confidence levels — Low, Neutral, and Confident.

This dataset was created as part of the ConfiDetect Tool, designed to evaluate human confidence through pose and body-language cues using computer vision.

🧩 Dataset Information

• Total Samples: 5,950
• Features: 18 (15 numeric + 3 categorical)
• Target Classes: Low, Neutral, High Confidence
• Format: CSV
• Normalization: Min–Max scaling applied
• Class Balance: 53% High, 28% Neutral, 19% Low

---

🧠 Methodology (Concise)

1. Data Collection:
   Real human video recorded for confidence behavior analysis.

2. Pose Extraction:
   MediaPipe Holistic model used to extract body and facial landmarks.

3. Feature Engineering:
   Computation of geometric ratios and angles such as eye-to-shoulder ratio, hip-shoulder difference, body lean, spine angle, and head tilt.

4. Label Assignment:
   Rule-based labeling (posture and facial cues) with human validation.

5. Normalization:
   All numeric attributes normalized for ML model compatibility.

---

📁 File Description

Column Name	Type	Description
eye_shoulder_y_ratio	float	Ratio between eye and shoulder Y-axis positions
shoulder_y_diff	float	Shoulder height difference
wrist_distance_x	float	Horizontal distance between wrists
wrist_shoulder_ratio	float	Wrist distance normalized by shoulder span
nose_eye_center_offset_x	float	Horizontal offset of nose from eye center
shoulder_span	float	Distance between left and right shoulders
hip_shoulder_y_diff	float	Difference between hip and shoulder alignment
body_lean_x	float	Forward/backward lean of upper body
shoulder_center_x	float	X-coordinate of shoulder midpoint
hip_center_x	float	X-coordinate of hip midpoint
spine_angle	float	Vertical spine inclination angle
eye_distance	float	Distance between eyes
head_tilt_angle	float	Head tilt angle in degrees
eye_distance_ratio	float	Normalized eye distance ratio
shoulder_slope	float	Slope between shoulders
head_direction	object	Head orientation (Left, Right, Straight)
arm_position	object	Arm posture (Open, Closed, Partially Open)
posture	object	Overall body posture description
confidence_label	object	Target class: Low, Neutral, High

---

🧮 Suggested Use

• Confidence Estimation Models
• Behavioral Analysis
• Affective Computing
• Body-Language Recognition
• Human-Computer Interaction Research

---

📊 Example Tasks

• Train classifiers (e.g., Random Forest, XGBoost, or SVM) to predict confidence.
• Build pose-based deep learning models with feature fusion.
• Evaluate correlations between physical posture and perceived confidence.

---

📜 Citation

Khubaib (2025). ConfiDetect-Confidence-Posture-Dataset: Pose-based Confidence Estimation using MediaPipe Features. Hugging Face.

---

🧾 License

MIT License — free for research and educational use.