MicroExpNet: An Extremely Small and Fast Model For Expression Recognition From Face Images

Cugu, Ilke
Sener, Eren
Akbaş, Emre
This paper is aimed at creating extremely small and fast convolutional neural networks (CNN) for the problem of facial expression recognition (FER) from frontal face images. To this end, we employed the popular knowledge distillation (KD) method and identified two major shortcomings with its use: 1) a fine-grained grid search is needed for tuning the temperature hyperparameter and 2) to find the optimal size-accuracy balance, one needs to search for the final network size (or the compression rate). On the other hand, KD is proved to be useful for model compression for the FER problem, and we discovered that its effects get more and more significant with decreasing model size. In addition, we hypothesized that translation invariance achieved using max-pooling layers would not be useful for the FER problem as the expressions are sensitive to small, pixelwise changes around the eye and the mouth. However, we have found an intriguing improvement in generalization when max-pooling is used. We conducted experiments on two widelyused FER datasets, CK+ and Oulu-CASIA. Our smallest model (MicroExpNet), obtained using knowledge distillation, is less than 1MB in size and works at 1851 frames per second on an Intel i7 CPU. Despite being less accurate than the state-of-the-art, MicroExpNet still provides significant insights for designing a microarchitecture for the FER problem.


Face classification with support vector machine
Kepenekci, B; Akar, Gözde (2004-04-30)
A new approach to feature based frontal face recognition with Gabor wavelets and support vector machines is presented in this paper. The feature points are automatically extracted using the local characteristics of each individual face. A kernel that computes the similarity between two feature vectors, is used to map the face features to a space with higher dimension. To find the identity of a test face, the possible labels of each feature vector of that face is found with support vector machines, then the ...
CEREBRA: A 3-D Visualization Tool for Brain Network Extracted from fMRI Data
Nasır, Barış; Yarman Vural, Fatoş Tunay (2016-08-20)
In this paper, we introduce a new tool, CEREBRA, to visualize the 3D network of human brain, extracted from the fMRI data. The tool aims to analyze the brain connectivity by representing the selected voxels as the nodes of the network. The edge weights among the voxels are estimated by considering the relationships among the voxel time series. The tool enables the researchers to observe the active brain regions and the interactions among them by using graph theoretic measures, such as, the edge weight and n...
Gürbüz, Yeti Ziya (2019-01-01)
An end-to-end trainable convolutional neural network (CNN) framework which mimics bag of visual words (BoVW) is proposed for image classification. To this end, a new paradigm for histogram-like image representation is introduced and optimal transport (OT) distance is utilized for the similarity assessment. Any patch of an image is considered as a unique visual word and the image is represented as the uniform histogram of the visual words with the histogram bins associated to embedding vectors according to t...
Occluded face recognition based on Gabor wavelets
Kepenekci, B; Tek, FB; Akar, Gözde (2002-09-25)
A new feature based approach to frontal face recognition with Gabor wavelets is presented in this paper. The feature points are automatically extracted using the local characteristics of each individual face in order to decrease the effect of occluded features. There is no training as in neural network approaches, thus single frontal face for each individual is enough as reference. Experimental results show that the proposed method achieves a recognition ratio of over %95.
Tubularity Tracking Based Automatic Road Detection from Sattelite Images
Gürbüz, Yeti Ziya; Alatan, Abdullah Aydın (2014-04-25)
In this paper, a novel approach based on tubularity tracking and graph cuts for road detection from satellites images is presented. The most important feature of the proposed method is its local peak detection filter. Unlike the tubularity based road or road like curvilinear structure detection methods presented in the literature, proposed method samples local peaks from tubularity image by tracking the peak points based on Bayesian filtering in order to construct graphs and introduces no significant comput...
Citation Formats
I. Cugu, E. Sener, and E. Akbaş, “MicroExpNet: An Extremely Small and Fast Model For Expression Recognition From Face Images,” 2019, Accessed: 00, 2020. [Online]. Available: https://hdl.handle.net/11511/35901.