You are on page 1of 3

Reference Technique Features Dataset Frame Performance

Dimension Evaluation
[1] Novel CNN Model. implemented 1 x 1 Fer2013, CK Different 65% on training and
20 layers of convolutions in our model and CK, datasets used 74% on validation
convolution operation in order to reduce the Chicago Face, so no specific
with ReLU and dimensionality before JAFFE, FEI dimensions
pooling undergoing. This helps face dataset,
reduce the multiplications IMFDB,
by a factor of 10, hence TFEID
reducing the computational
complexity of the network.
[2] Transfer training Novel Dataset 227 X 227 Testing Accuracy
method and AlexNet's we created a convolutional 74%
pre-trained networks neural network and trained
it over our dataset by
scratching method and we
achieved 50% accuracy rate.
Then, we increased the
number of images in our
database by 3 times, and get
better accuracy which is
62%
[3] CNN is trained with two-layer convolution Novel Dataset 32 x 32 78.04%
the emotion image network model
dataset, utilizing
Adam as the optimizer
and the categorical
cross-entropy as the
loss function
[4] The emotion detection We find different patterns of 48 × 48 65%
module uses the single or combined Trained using
trained CNN classifier emotions of a learner during FER2013
to the learning session.
detect the different b) We propose a definite Tested on
emotions in the image. model to assess the state of Feature
The state of mind mind Extraction and
identification module of a learner from his/her analysis from
takes a sequence of 6 different emotional captured
images as a window expressions images
frame to identify the at any point of time in a And
emotion pattern of the learning session. Sensor
learner for the last 6 analyzed data
seconds. The learner‟s
state of mind is
identified from the
derived
emotion pattern.
Further, the identified
state of mind is sent to
LMS as feedback of
learner‟s emotion.
Based on the identified
state of mind of a
learner, the LMS
adapts the
instructional or
pedagogical method
in real-time, making
more suitable for the
particular learner.
[5] GoogLeNet and By using the whole dataset FER-2013 48 × 48 85%
Alexnet two as mentioned, an extra dataset
popular and wide used complexity of feature
deep learning methods recognition is introduced,
and GoogLeNet might be
more
efficient after many
iterations when AlexNet
will be overtrained, due to
its lack of
depth, but in the first 5000
iterations, thanks to its
shallow architecture,
AlexNet
seems to reach and get
ahead of GoogLeNet’s
accuracy as it manages to
train quite
quicker.
[6] For face detection a fast and 48x48 69%
deep learning robust face detector FER2013
(similar to VGG-Net) implemented in OpenCV dataset
is used has been used. Given a real-
time live stream, the face
The CONV layers in detector examines each
the network are location of the image and
3x3(filter size). classifies it as a face or no
face region. To decide how
2. The number of to classify each image
filters learned by each location the classifier uses
CONV layer is data stored in an XML file.
doubled as we go For XML data detection,
deeper in the network. OpenCV's Haar Cascade
3. The CONV and FC classifier file is chosen.
layers are initialized
using MSRA/He et al.
method. This works
better for the VGG
family of networks
and enables the
network to learn
faster.
4. ELU is used instead
of ReLU to boost
classification
accuracy.
[1] Jaiswal, Shruti, and Gora Chand Nandi. "Robust real-time emotion detection system
using cnn architecture." Neural Computing and Applications 32.15 (2020): 11253-11262.
[2] E. Dandıl and R. Özdemir, “Real-time Facial Emotion Classification Using Deep
Learning”, DataSCI, vol. 2, no. 1, pp. 13-17, Jul. 2019.
[3] E. Pranav, S. Kamal, C. Satheesh Chandran and M. H. Supriya, "Facial Emotion
Recognition Using Deep Convolutional Neural Network," 2020 6th International
Conference on Advanced Computing and Communication Systems (ICACCS), 2020, pp.
317-320, doi: 10.1109/ICACCS48705.2020.9074302.
[4] Mukhopadhyay, Moutan, et al. "Facial emotion detection to assess Learner's State of
mind in an online learning system." Proceedings of the 2020 5th International Conference
on Intelligent Information Technology. 2020.
[5] Giannopoulos, Panagiotis, Isidoros Perikos, and Ioannis Hatzilygeroudis. "Deep learning
approaches for facial emotion recognition: A case study on FER-2013." Advances in
hybridization of intelligent methods. Springer, Cham, 2018. 1-16.
[6] Sinha, Avigyan, and R. P. Aneesh. "Real Time Facial Emotion Recognition using Deep
Learning." International Journal of Innovations and Implementations in Engineering 1
(2019).

You might also like