DEEP LEARNING LAB

Paper Code: 
24DCAI 802A
Credits: 
06
Periods/week: 
12
Max. Marks: 
100.00
Objective: 

Course Objectives:

The course will enable the students to

  1. Demonstrate the concepts of computer vision and image processing.
  2. Apply an appropriate deep learning model for a given real world problem.
  3. Optimize the models using various optimization algorithms and fine tuning the     hyper parameters.

 

Course Outcomes: 

Course

Learning outcome

(at course level)

Learning and teaching strategies

Assessment Strategies

Course Code

Course

title

 

24DCAI 802A

 

DEEP LEARNING LAB

(PRACTICAL)

CO157.Apply the concepts of image processing.

CO158.  Compare and evaluate the performance of different deep learning models.

CO159. Develop deep learning models to encode and reconstruct the original data.

CO160. Modify the hyper-parameters and optimize deep learning models.

CO161. Develop custom ensembles using bagging, boosting and stacking.

CO162. Contribute effectively in course-specific interaction

 

Approach in teaching.

Interactive Lectures, Group Discussion,

Case Study

 

Learning activities for the students.

Self-learning assignments, Machine Learning exercises, presentations

 

Class test, Semester end examinations, Quiz, Practical Assignments, Presentation

 

Practical based on following topics.

Computer Vision: -

  1. Simulation and Display of an Image, Negative of an Image (Binary & Gray Scale)
  2. Implementation of Relationships between Pixels
  3. Implementation of Transformations of an Image
  4. Contrast stretching of a low contrast image, Histogram, and Histogram Equalization.
  5. Display of bit planes of an Image
  6. Display of FFT(1-D & 2-D) of an image
  7. Computation of Mean, Standard Deviation, Correlation coefficient of the given Image.
  8. Implementation of Image Smoothening Filters(Mean and Median filtering of an Image)
  9. Implementation of image sharpening filters and Edge Detection using Gradient Filters
  10. Image Compression by DCT,DPCM, HUFFMAN coding
  11. Image segmentation
  12. Image recognition

 

Deep Learning: -

  1. Design and implement a CNN for Image Classification
    1. Select a suitable image classification dataset (medical imaging, agricultural, etc.).
    2. Optimized with different hyper-parameters including learning rate, filter size, no. of layers, optimizers, dropouts, etc.
  2. Design RNN

a) Select a suitable time series dataset. Example – predict sentiments based on product reviews

b) Apply for prediction

3. Implement an artificial neural network on GPUs

a) Implement ANN on GPUs.

b) Deploy the model using Amazon SageMaker or other platforms available.

4. Implement MLP, RBM, DBN algorithms.

5. Implement Auto-encoders for the following:

a) Data Compression

b) Image de-noising

c) Dimensionality reduction

6. Design and evaluate ensemble models using boosting stacking and bagging.

 

ESSENTIAL READINGS: 

Suggested Text Books:

  1. Zaccone, G., Karim, M. R., Menshawy, A. "Deep Learning with TensorFlow: Explore neural networks with Python", Packt Publisher, 2017.
  2. Sebastian Raschka & Vahid Mirjalili,” Python Machine Learning”, Second Edition,Packt>.

Aston Zhang, Zachary C. Lipton, Mu Li, and Alexander J. Smola, "Dive into Deep Learning", Amazon Science, 2021.

REFERENCES: 

Suggested Reference Books:

  1. Goodfellow I., BengioY., and Courville A., "Deep Learning", MIT Press, 2016, ISBN: 978-0262035613.
  2. McKinney, Python for Data Analysis. O’ Reilly Publication,2017.

 

Reference Journals:

  1. Journal of Real-Time Image Processing (JRTIP) springer: https://www.springer.com/journal/11554
  2. IEEE Transactions on Image Processing:https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=83

 

e-Resources including links

  1. NPTEL: Deep Learning for Computer Vision, IIT Hyderabad : https://nptel.ac.in/courses/106106224
  2. Image processing by Saikiran Panjala, Slideshare: https://www.slideshare.net/MadhushreeGhosh3/image-processing-76619758

 

Academic Year: 
2024-25
  • Printer-friendly version