Efficientnet-Lite and Hybrid CNN-KNN Implementation for Facial Expression Recognition on Raspberry Pi

被引:39
作者
Ab Wahab, Mohd Nadhir [1 ]
Nazir, Amril [2 ]
Ren, Anthony Tan Zhen [1 ]
Noor, Mohd Halim Mohd [1 ]
Akbar, Muhammad Firdaus [3 ]
Mohamed, Ahmad Sufril Azlan [1 ]
机构
[1] Univ Sains Malaysia, Sch Comp Sci, Minden 11800, Penang, Malaysia
[2] Zayed Univ, Coll Technol Innovat, Dept Informat Syst & Technol Management, Abu Dhabi, U Arab Emirates
[3] Univ Sains Malaysia, Sch Elect & Elect Engn, Engn Campus, Nibong Tebal 14300, Penang, Malaysia
关键词
Convolutional neural networks; Hidden Markov models; Feature extraction; Support vector machines; Training; Computational modeling; Real-time systems; EfficientNet-Lite; hybrid CNN-KNN; facial expression recognition; Raspberry Pi; emotion recognition;
D O I
10.1109/ACCESS.2021.3113337
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Facial expression recognition (FER) is the task of determining a person's current emotion. It plays an important role in healthcare, marketing, and counselling. With the advancement in deep learning algorithms like Convolutional Neural Network (CNN), the system's accuracy is improving. A hybrid CNN and k-Nearest Neighbour (KNN) model can improve FER's accuracy. This paper presents a hybrid CNN-KNN model for FER on the Raspberry Pi 4, where we use CNN for feature extraction. Subsequently, the KNN performs expression recognition. We use the transfer learning technique to build our system with an EfficientNet-Lite model. The hybrid model we propose replaces the Softmax layer in the EfficientNet with the KNN. We train our model using the FER-2013 dataset and compare its performance with different architectures trained on the same dataset. We perform optimization on the Fully Connected layer, loss function, loss optimizer, optimizer learning rate, class weights, and KNN distance function with the k-value. Despite running on the Raspberry Pi hardware with very limited processing power, low memory capacity, and small storage capacity, our proposed model achieves a similar accuracy of 75.26% (with a slight improvement of 0.06%) to the state-of-the-art's Ensemble of 8 CNN model.
引用
收藏
页码:134065 / 134080
页数:16
相关论文
共 27 条
  • [1] [Anonymous], 1978, FACIAL ACTION CODING
  • [2] [Anonymous], FER 2013 DATASET
  • [3] Carlson N., 2012, PHYSIOL BEHAV
  • [4] Insights Into Efficient k-Nearest Neighbor Classification With Convolutional Neural Codes
    Gallego, Antonio-Javier
    Calvo-Zaragoza, Jorge
    Ramon Rico-Juan, Juan
    [J]. IEEE ACCESS, 2020, 8 : 99312 - 99326
  • [5] Improving Convolutional Neural Networks' Accuracy in Noisy Environments Using k-Nearest Neighbors
    Gallego, Antonio-Javier
    Pertusa, Antonio
    Calvo-Zaragoza, Jorge
    [J]. APPLIED SCIENCES-BASEL, 2018, 8 (11):
  • [6] Goodfellow Ian J., 2013, Neural Information Processing. 20th International Conference, ICONIP 2013. Proceedings: LNCS 8228, P117, DOI 10.1007/978-3-642-42051-1_16
  • [7] Izard Carroll E, 1991, The Psychology of Emotions
  • [8] Robust Facial Expression Recognition Based on Local Directional Pattern
    Jabid, Taskeed
    Kabir, Md. Hasanul
    Chae, Oksam
    [J]. ETRI JOURNAL, 2010, 32 (05) : 784 - 794
  • [9] Kimmel R, 2017, ARXIV PREPRINT ARXIV
  • [10] Loza-Alvarez A, 2018, 2018 XX CONGRESO MEXICANO DE ROBOTICA (COMROB)