Research Article
Facial Landmark Detection Using Deep Learning: A Comprehensive Approach with Resnet18
@INPROCEEDINGS{10.4108/eai.28-4-2025.2358028, author={Sk. Mastan Sharif and N. Sri Harsha and Sk. Abdul Subhan and Y. Threeshal}, title={Facial Landmark Detection Using Deep Learning: A Comprehensive Approach with Resnet18}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part II}, publisher={EAI}, proceedings_a={ICITSM PART II}, year={2025}, month={10}, keywords={resnet18 ibug 300w dataset cnn (convolutional neural network)}, doi={10.4108/eai.28-4-2025.2358028} }- Sk. Mastan Sharif
N. Sri Harsha
Sk. Abdul Subhan
Y. Threeshal
Year: 2025
Facial Landmark Detection Using Deep Learning: A Comprehensive Approach with Resnet18
ICITSM PART II
EAI
DOI: 10.4108/eai.28-4-2025.2358028
Abstract
Facial landmark detection is a crucial computer vision problem and has a number of applications in facial recognition, emotion detection, and augmented reality. A deep learning-based approach for detecting facial landmarks from a ResNet18-based convolutional neural network (CNN) is discussed in this paper. The model, which is trained and validated with the iBUG 300-W dataset where facial landmarks are annotated, detects facial landmarks efficiently. During training, several operations such as rotation, cropping, resizing, and color jittering are performed on data to increase the generalization power of the model. Model performance is assessed considering observing training and validation loss values over multiple epochs. From the results, we can see that the proposed method is capable enough to detect facial landmarks precisely, even 68 landmarks. Considering training and validation loss trends for preventing over- fitting and increasing model performance are also explained in the paper.
