Datasets
Standard Dataset
Shape Predictor 68 Face Landmarks Model for Eye-Tracking-Based Cursor Control
- Citation Author(s):
- Submitted by:
- Jaswanth Nalamaru
- Last updated:
- Thu, 02/20/2025 - 22:55
- DOI:
- 10.21227/mjjq-f259
- Data Format:
- License:
9 Views- Categories:
- Keywords:
Abstract
This dataset supports the LookCursor AI project, which implements eye-tracking-based cursor control using OpenCV and Dlib. The primary file included is shape_predictor_68_face_landmarks.dat, a pre-trained model used to detect and map 68 facial landmarks essential for tracking eye movements. The dataset enables accurate facial feature detection, which is critical for cursor movement based on eye gaze. This resource is valuable for researchers working on assistive technology, human-computer interaction (HCI), and computer vision applications. The dataset is used in conjunction with Python-based scripts to enhance real-time eye-tracking accuracy.
This dataset contains a pretrained model file (shape_predictor_68_face_landmarks.dat) and two Python scripts (lookcursorai.py, utils.py). The dataset enables eye-tracking-based cursor control using Dlib, OpenCV, and PyAutoGUI.
How to Use:
1.Install Dependencies:
pip install dlib opencv-python numpy pyautogui
2.Run the Script:
python lookcursorai.py
3.The system will:
- Detect facial landmarks in real time.
- Track eye movements.
- Move the cursor based on gaze direction.
Files Included:
- shape_predictor_68_face_landmarks.dat – Pretrained facial landmark model.
- lookcursorai.py – Main script for gaze-based cursor control.
- utils.py – Helper functions for facial landmark processing.
This dataset is useful for AI researchers, HCI developers, and accessibility solutions.
Dataset Files
- shape_predictor_68_face_landmarks.dat (95.08 MB)
- lookcursorai.py (6.41 kB)
- utils.py (1.53 kB)
Documentation
| Attachment | Size |
|---|---|
 Tiltle.pdf | 330.81 KB |
