Đồ án tốt nghiệp - Facial Landmark detection on multiple platforms using quantized mobilenetv2 FEAA

MINISTRY OF EDUCATION AND TRAINING
HCMC UNIVERSITY OF TECHNOLOGY AND EDUCATION
--

GRADUATION THESIS
MAJOR: COMPUTER ENGINEERING TECHNOLOGY

FACIAL LANDMARK DETECTION ON MULTIPLE
PLATFORMS USING QUANTIZED MOBILENETV2
FOR EDGE AI APPLICATIONS

INSTRUCTOR

PHAM VAN KHOA

STUDENT

DO MANH DUNG
DANG TRUNG NGHIA

HO CHI MINH CITY – 07/2025

HCMC UNIVERSITY OF TECHNOLOGY AND EDUCATION
FACULTY OF INTERNATIONAL EDUCATION
--

GRADUATION PROJECT

FACIAL LANDMARK DETECTION ON
MULTIPLE PLATFORMS USING QUANTIZED
MOBILENETV2 FOR EDGE AI APPLICATIONS
ĐỖ MẠNH DŨNG
Student ID: 21119301
ĐẶNG TRUNG NGHĨA
Student ID: 21119313
Major: COMPUTER ENGINEERING TECHNOLOGY
Advisor: PHẠM VĂN KHOA, PhD.

HO CHI MINH CITY – 07/2025

HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY AND EDUCATION
FACULTY OF INTERNATIONAL EDUCATION

GRADUATION PROJECT

FACIAL LANDMARK DETECTION ON MULTIPLE
PLATFORMS USING QUANTIZED MOBILENETV2
FOR EDGE AI APPLICATIONS
ĐỖ MẠNH DŨNG
Student ID: 21119301
ĐẶNG TRUNG NGHĨA
Student ID: 21119313
Major: COMPUTER ENGINEERING TECHNOLOGY
Advisor: PHẠM VĂN KHOA, PhD.

HO CHI MINH CITY – 07/2025

THE SOCIALIST REPUBLIC OF VIETNAM
Independence – Freedom– Happiness

-Ho Chi Minh City, July …, 2025

GRADUATION PROJECT ASSIGNMENT
Student name: Đỗ Mạnh Dũng

Student ID: 21119301

Student name: Đặng Trung Nghĩa

Student ID: 21119313

Major: Computer Engineering Technology

Class: 21119FIE

Advisor: Phạm Văn Khoa, PhD.

Phone number: 0397079864

Date of assignment: _____________________ Date of submission: _____________
1. Project title: Facial Landmark Detection on Multiple Platforms using Quantized MobileNetV2
for Edge AI Applications
2. Initial materials provided by the advisor:
3. Content of the project:
-

Build and train a lightweight facial landmark detection model using MobileNetV2
architecture and the WFLW dataset.

-

Integrate a pose-estimation auxiliary branch and train the model with a weighted poseand class-aware loss function.

-

Perform Post-Training Quantization (PTQ) and Quantization-Aware Training (QAT) to
compress the model to INT8 format using Vitis AI.

-

Deploy and evaluate the quantized model on multiple platforms (CPU, GPU, Raspberry
Pi, Jetson Nano, FPGA KV260) for real-time performance and energy efficiency.

4. Final product: ________________________________________________________

CHAIR OF THE PROGRAM

ADVISOR

(Sign with full name)

(Sign with full name)

THE SOCIALIST REPUBLIC OF VIETNAM
Ind
...

...
se, J. (2008). “FPGA architecture: Survey and challenges.
Foundations and Trends® in Electronic Design Automation”, 2(2), 135-253.
[35]: Nurvitadhi, E., Venkatesh, G., Sim, J., Marr, D., Huang, R, Ong Gee Hock, J, ... &amp
Deborah, M. (2017). “Can FPGAs beat GPUs in accelerating next-generation deep
learning?” In Proceedings of the 2017 ACM/SIGDA International Symposium on FieldProgrammable Gate Arrays (pp. 5-14).
[36]: Betz, V., Rose, J., &amp Marquardt, A. (1999). “Architecture and CAD for deepsubmicron FPGAs. Springer Science &amp Business Media”.
[37] Zhang, C., Li, P., Sun, G., Guan, Y., Xiao, B., &amp Cong, J. (2015). “Optimizing FPGAbased accelerator design for deep convolutional neural networks”. In Proceedings of the
2015 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (pp.
161-170).
[38]: Lemieux, G. G., &amp Lewis, D. (2004). “Design of interconnection networks for
programmable logic. Springer Science &amp Business Media”.
[39]: Farabet, C., Martini, B., Akselrod, P., Talay, S., LeCun, Y., &amp Culurciello, E. (2010).
“Hardware accelerated convolutional neural networks for synthetic vision systems. In
Proceedings of 2010 IEEE International Symposium on Circuits and Systems (pp. 257260)”.
[40]: U. Farooq et al, “Tree-Based Heterogeneous FPGA Architectures , FPGA
Architectures: An Overview”, 2012
[41]: Ketkar, N., &amp Moolayil, J. “Deep learning with Python: Learn best practices of deep
learning models with PyTorch. Apress”, 2021.
71

[42]: Guo, X., Li, S., Yu, J., Zhang, J., Ma, J., Ma, L., Liu, W., &amp Ling, H. (2019). “PFLD:
A Practical Facial Landmark Detector [Preprint]. arXiv:1902.10859 v2”.
[43]: Xiang Wang, Kai Wang, Shiguo Lian, “A Survey on Face Data Augmentation”,
CloudMinds Technologies Inc, 2019.
[44]: A. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto,
and H. Adam. “Mobilenets: Efficient convolutional neural net works for mobile vision
applications. CoRR, abs/1704.04861”, 2017.
[45]: W. Wu, C. Qian, S. Yang, Q. Wang, Y. Cai, and Q. Zhou, “Look at boundary: A
boundary-aware face alignment algorithm. In CVPR”, 2018.
[46]: A. Kumar and R. Chellappa, “Disentangling 3d pose in a dendritic cnn for
unconstrained 2d face alignment. In CVPR”, 2018
[47]: A. Jourabloo, X. Liu, M. Ye, and L. Ren, “Pose invariant face alignment with a single
cnn. In ICCV”, 2017.
[48]: H. Yang, W. Mou, Y. Zhang, I. Patras, H. Gunes, and P. Robinson, “Face alignment
assisted by head pose estimation. In BMVC”, 2015.

72