Writing a thesis on Facial Expression Recognition can be an incredibly challenging endeavor.

It
requires a deep understanding of various concepts in computer vision, machine learning, and
psychology, among other fields. From conducting extensive literature reviews to designing and
implementing complex algorithms, the process demands a significant amount of time, effort, and
expertise.

One of the most daunting aspects of writing a thesis in this field is the need for meticulous data
collection and preprocessing. Gathering high-quality facial expression datasets can be difficult, and
cleaning and annotating the data is a time-consuming task. Moreover, ensuring the reliability and
validity of the collected data adds another layer of complexity to the research process.

Another challenge lies in developing robust algorithms for facial expression recognition. This
involves selecting appropriate features, choosing suitable machine learning models, and fine-tuning
parameters to achieve optimal performance. Additionally, evaluating the effectiveness of these
algorithms requires rigorous testing methodologies and statistical analysis.

Furthermore, writing a thesis involves not only conducting research but also effectively
communicating the findings. This includes structuring the thesis logically, presenting results clearly,
and discussing the implications of the research findings. Additionally, adhering to academic writing
conventions and formatting guidelines adds another level of difficulty to the process.

Given the complexity and demanding nature of writing a thesis on Facial Expression Recognition,
seeking professional assistance can be invaluable. ⇒ HelpWriting.net ⇔ offers expert thesis
writing services to support students throughout their academic journey. With a team of experienced
researchers and writers, ⇒ HelpWriting.net ⇔ can provide personalized assistance tailored to your
specific needs. From literature review and data analysis to thesis writing and editing, they offer
comprehensive support to ensure the success of your research project. Save yourself the stress and
hassle of writing a thesis alone – trust ⇒ HelpWriting.net ⇔ to guide you through the process and
deliver exceptional results.
Facial Expression Recognition Using Computer Vision: A Systematic Review. Appl. Sci. 2019, 9,
4678. In response to emotional stimuli, rapid muscle movements happen in the onset phase, which is
involuntary and shows genuine emotional leakage. As for the feature extraction step, they combined
the grayscale face image with its corresponding basic LBP and mean LBP feature maps to form a
three-channel input. Sometimes the facial expression analysis has been confused with emotion
analysis in the computer vision domain. This part alone achieved 39.95% accuracy on the AFEW
database. We showed that facial micro-expressions could be combined with other modalities in
emotion recognition. The first perspective is the well-known discrete emotion model introduced by
Ekman and Friesen (1971) which categorized emotions into six basic types; happiness, sadness,
surprise, anger, disgust, and fear. Haar features are used in this algorithm and they are applied in all
training images to find the best threshold which will classify the faces as positive or negative
detections. These parts were then combined with weights for the final classification. See Full PDF
Download PDF About Press Blog People Papers Topics Job Board We're Hiring. Demonstrated
CNN features used by LSTM for sequence learning; SoftMax is used. Previous Article in Journal A
Systematic Procedure for Utilization of Product Usage Information in Product Development.
Applications of facial expression recognition system include border security systems, forensics,
virtual reality, computer games, robotics, machine vision, video conferencing, user profiling for
customer satisfaction, broadcasting and web services. The hybrid techniques are, thus, equipped to
distinguish emotions from image sequences. The feature vectors are combined and reduced and then
applied to SVM classifier for training process. These models were fused into a novel integration
scheme to increase FER efficiency. NTSC cameras record images at 30 frames per second, The
implications of down-sampling from this rate are unknown. Kreibig (2010) showed that although
EDA signals show changes in emotional arousal, more research is needed to identify the type of
emotion using EDA signals. By smoothing an image, one can capture relevant patterns while
filtering the noise. Facial Expression Recognition Using Computer Vision: A Systematic Review.
Comparisons of deep-learning-based approaches for FER. Precision: This indicates how accurate or
precise a model is, and it yields the actual positive values from a set of predictions by any algorithm.
We identified each trial's region of interest (ROI) using a landmark-based spotting strategy for
detecting micro-expressions. They achieved an accuracy of 87.3% for EEG and 99.8% for facial
micro-expression from their experimental dataset. Compared to the previous work reported in Table
1, the accuracy of the proposed methods is considerably high while considering the subject-
independent approach, which is the most challenging evaluation condition. This study of recognizing
facial expression is one of the challenging research areas in image analysis and computer vision.
Ganapathy et al. (2021) showed that Multiscale Convolutional Neural Networks (MSCNN) are
effective in extracting deep features of GSR signals and classifying emotions. Furthermore, the
subjects of some FER databases were asked to pose certain emotions towards a reference, while
others tried to stimulate spontaneous and genuine facial expressions. However, later studies showed
improvement by combining EEG and facial expressions. All models were trained on a computer with
Gnu-Linux Ubuntu 18.04, Intel(R) Core(TM) i7-8700K CPU (3.70 GHz) with six cores.
Editors select a small number of articles recently published in the journal that they believe will be
particularly. A CNN architecture’s last FC layer calculates the class probability of an entire input
image. In the feature extraction step, they extracted AUs and used Supervised Descent Method
(SDM) to track them. Our methods were evaluated based on a subject-independent approach.
Recognizing emotions is becoming more critical in a wide range of application domains such as
healthcare, education, human-computer interaction, Virtual Reality, intelligent agents, entertainment,
and more. In the end, we present the results, limitations, and future works. We left the participant
alone in the room during the experiment to prevent any distraction or psychological effects of a
stranger's presence. However, there is a common discrepancy of accuracy when testing in controlled
environment databases and in wild environment databases. We believe that the peak time of feeling
emotions with the most intensity is affected by many factors such as the stimuli flow, participant
personality, or previous experiences. Hence, the imbalance state among the training and test sets for
each set depended on the participants' rating. Further, different FER datasets for evaluation metrics
that are publicly available are discussed and compared with benchmark results. The recent
popularization of Machine Learning made an obvious breakthrough in the research field. For
example, although facial expressions can convey emotion, they can also express intention, cognitive
processes, physical effort, or other intra- or interpersonal meanings. In natural environments, multiple
people interacting with each other are likely, and their effects need to be understood. As for the pre-
processing step, they applied SDM to track facial features, face frontalization, rescaling, and applied
a DCT based method for intensity normalization. Find support for a specific problem in the support
section of our website. Paper should be a substantial original Article that involves several techniques
or approaches, provides an outlook for. The convention ML approach consists of face detection,
feature extraction from detected faces and emotion classification based on extracted features. I plan
to post updates here on my progress and things I am working on, both for my own benefit and
tracking and also in the hopes that someone else may find my work interesting. The feature vectors
are combined and reduced and then applied to SVM classifier for training process. Methods that
work for intense expressions may generalize poorly to ones of low intensity. Since EEG signals are
sensitive to muscle artifacts ( Jiang et al., 2019 ), these kinds of datasets used passive tasks like
watching videos or listening to music to minimize the subject movements. It is able to capture spatial
information of frequency, position, and orientation from an image, and extract subtle local
transformations effectively. Depending on the value of these metrics, one can predict which emotion
certain subject is feeling. Most previous works evaluate their methods subject dependently or cross-
subject when there are some trials of all participants in the training and test sets. The cameras are
moving together with the head to eliminate the scale and orientation variance of the acquired face
images. In Table 6, the last column shows the accuracy obtained for each work in predicting emotions
using the circumplex model. The extracted facial features are represented by three features vectors:
the Zernike moments, LBP features and DCT transform components. MMI Database. Available
online: (accessed on 8 September 2019). However, their results concluded that a multimodal
approach is generally better than a unimodal approach.
Since spotting methods still need more exploration and are an open challenge ( Oh et al., 2018 ), we
could improve our results in the future by manually annotating the DEAP and our dataset. Unless
they are tested at a range of sampling rates, the robustness to sampling rate and resolution cannot be
assessed. With image data from a single camera, out-of-plane rotation may be difficult to
standardize. This method has the advantage of improving the contrast in an image, highlighting
facial features and reducing the interference caused by different lighting conditions. Based on the
EmotiW Challenge results, the future direction for FER in uncontrolled environments seems to be
converging into: Pre-processing techniques that normalize pose-variant faces as well as the image
intensity. In order to reduce the dimensionality, a subset feature selection algorithm is used prior to
training process. This work emphasized the importance of pre-processing the data by testing their
system with the original video frames, obtaining an average accuracy of just 20%. They used
feature-level and decision-level fusion strategies. However, the latest comparison of different
approaches is presented in Table 3. This work won the EmotiW 2018 challenge with the best
submission achieving 61.87% accuracy. Figure 23 illustrates the EmotiW Challenge winners’
accuracy over time. Some used an end-to-end CNN approach, which is only fed with ROIs (faces)
that could have been pre-processed or not. An input image is run through several hidden layers of the
CNN that will decompose it into features. This time was 33 min for our dataset (79 s for each
epoch) because of each participant's lower number of trials. This method can be effectively
implemented in an FER system since, whenever one goes from a neutral expression to a peak facial
expression, there is an obvious motion in the face that can be estimated. They also used a capsule
network to find the relationship between channels. Table 6 sums up results of works that approached
this problem by analyzing the circumplex model. Generally, an FER system consists of the following
steps: image acquisition, pre-processing, feature extraction, classification, or regression, as shown in
Figure 1. However, in this dataset, only 22 participants have video data, and for 4 of them, some
trials have been missed. They represent a complete set of basic facial actions, allowing the
representation of facial expressions. The videos were captured using a powerful 100 frame per
second camera. Training process of CNN model for facial emotion recognition. The EmotiW
Challenge is divided into two different competitions: one is based in a static approach, using the
SFEW database, and the other one is based in an audiovisual approach, using the AFEW database.
Figure 4 shows some examples of the introduced FER databases: Table 1 sums up these introduced
databases and provides their website. 4. Pre-Processing Pre-processing is one of the most important
phases, not only in FER systems, but in any Machine Learning based system. LSTM takes
appearance features extracted by the CNN over individual video frames as input and encodes
motion later, while C3D models appearance and motion of video simultaneously. Also, the F-Score
of the ROI-based LSTM is relatively close to or sometimes better than using LSTM on the whole of
the data. It has the advantage of reducing overfitting over just one DT, since it reduces the bias by
averaging the predictions of the ensemble. Then we analyze the EEG and physiological data around
the emergence of micro-expressions in each trial in comparison to the analysis of the entire trial. If
the neighbor pixel value is greater than or equal to the center pixel value, then it takes the value “1”,
else it takes the value “0”. Multiple requests from the same IP address are counted as one view. For
facial classification, they used the K-Nearest-Neighbor (KNN) classifier for EEG and Support Vector
Machine (SVM).