ntroduction: Coronavirus has become the center of attention; most people worldwide are suffering greatly, and millions o
Posted: Tue May 24, 2022 7:48 am
ntroduction: Coronavirus has become the center of attention;
most people worldwide are
suffering greatly, and millions of people are dying every day
because of COVID-19. In
combination with other safety precautions such as regular hand
hygiene and social isolation,
face masks help restrict the spread of the coronavirus, according
to the World Health
Organization (WHO). As you may be aware, the WHO has suggested that
even healthy
individuals wear masks when stepping outside their houses into
areas were maintaining a safe
distance from other people is difficult.
Assume that a local housing society has begun to capture images at
the entrance/exit gates
and public areas within and around the residential building to
recognize the people wearing
a mask from the specific group of people who refuse to comply.
Assume you are in charge of
classifying images and developing a Deep Learning model to
recognize face masks.
Dataset:
The collection contains about 12k images, nearly 328.92MB in size.
All the photos with the
face mask (6K) were scraped from Google, and all the images without
the face mask were pre-
processed using Jessica Li's Celebrity Face dataset
(https://www.kaggle.com/jessicali9530).
Dataset will be provided in zip files.
Task1: Import Libraries:
a. Import required libraries (required)
Task2: Dataset
a. Load dataset
b. Rescale images.
c. Split dataset into training (80%) and validation (20%)
Task3: Build model to detect mask and unmask.
a. Choose suitable model from various Convolutional Neural
Network.
b. Built model from scratch or use pre-trained models with few
added layers and
parameters.
With-Mask Without-Mask
Task4: Compile the Model
a. A few more parameters are required before the model can be used
for training.
These are added at the build step of the model: Choose these values
and explain
why you choose that value.
a.1 Loss function
a.2 learning rate
a.3 Optimizer
a.4 Metrics
Task 5: Train and Test the model.
a. Pass List of callbacks (e.g., early stopping)
b. Customize No. of epoch, batch size (Choose these values and
explain why you
choose that value).
c. Feed the training data to the built model.
d. Ask model to make predictions on test set.
e. Verify that the predictions match the labels from the test
labels.
Prepare a report: (1500-2000 words)
Your report should contain the following:
1) Abstract: Provide a concise summary of project (cover brief
introduction to the topic,
explaining why the topic is important, indication of your methods
and approach, and
results you obtained). [note: There should be an introduction, a
body, and a
conclusion. It should be a well-developed paragraph with precise
terminology that is
understandable to a large audience.]
2) Introduction and Literature Review (In literature review provide
comprehensive
summary of previous work).
3) The methods applied for solving each task, describe architecture
used and reason for
choosing that architecture.
4) Results: take a screenshot of the results (e.g., learning
graphs) and attach it in
document.
5) Challenges and problem during project
6) References: cite all the information used in this report.
most people worldwide are
suffering greatly, and millions of people are dying every day
because of COVID-19. In
combination with other safety precautions such as regular hand
hygiene and social isolation,
face masks help restrict the spread of the coronavirus, according
to the World Health
Organization (WHO). As you may be aware, the WHO has suggested that
even healthy
individuals wear masks when stepping outside their houses into
areas were maintaining a safe
distance from other people is difficult.
Assume that a local housing society has begun to capture images at
the entrance/exit gates
and public areas within and around the residential building to
recognize the people wearing
a mask from the specific group of people who refuse to comply.
Assume you are in charge of
classifying images and developing a Deep Learning model to
recognize face masks.
Dataset:
The collection contains about 12k images, nearly 328.92MB in size.
All the photos with the
face mask (6K) were scraped from Google, and all the images without
the face mask were pre-
processed using Jessica Li's Celebrity Face dataset
(https://www.kaggle.com/jessicali9530).
Dataset will be provided in zip files.
Task1: Import Libraries:
a. Import required libraries (required)
Task2: Dataset
a. Load dataset
b. Rescale images.
c. Split dataset into training (80%) and validation (20%)
Task3: Build model to detect mask and unmask.
a. Choose suitable model from various Convolutional Neural
Network.
b. Built model from scratch or use pre-trained models with few
added layers and
parameters.
With-Mask Without-Mask
Task4: Compile the Model
a. A few more parameters are required before the model can be used
for training.
These are added at the build step of the model: Choose these values
and explain
why you choose that value.
a.1 Loss function
a.2 learning rate
a.3 Optimizer
a.4 Metrics
Task 5: Train and Test the model.
a. Pass List of callbacks (e.g., early stopping)
b. Customize No. of epoch, batch size (Choose these values and
explain why you
choose that value).
c. Feed the training data to the built model.
d. Ask model to make predictions on test set.
e. Verify that the predictions match the labels from the test
labels.
Prepare a report: (1500-2000 words)
Your report should contain the following:
1) Abstract: Provide a concise summary of project (cover brief
introduction to the topic,
explaining why the topic is important, indication of your methods
and approach, and
results you obtained). [note: There should be an introduction, a
body, and a
conclusion. It should be a well-developed paragraph with precise
terminology that is
understandable to a large audience.]
2) Introduction and Literature Review (In literature review provide
comprehensive
summary of previous work).
3) The methods applied for solving each task, describe architecture
used and reason for
choosing that architecture.
4) Results: take a screenshot of the results (e.g., learning
graphs) and attach it in
document.
5) Challenges and problem during project
6) References: cite all the information used in this report.