Smashing September Sale - Get 25% Off Certificates and Diplomas! Limited-time Offer - ends Friday, 17th September 2021

Claim My 25% OFF

Data Processing Using Convolutional and Recurrent Neural Networks

In this free online course, learn about the methods of processing images and sequential data using CNNs and RNNs.

Publisher: NPTEL
This free online course explains the process of identifying components of an image using convolutional and recurrent neural networks. Explore how these neural networks function in processing temporal and spatial information, and the methods used in distinguishing unexpected items or events in images that differ from a pattern.
Data Processing Using Convolutional and Recurrent Neural Networks
  • Duration

    5-6 Hours
  • Students

  • Accreditation






View course modules


What are convolutional neural networks and how do they work? Convolutional neural networks (CNNs) are among the most common types of neural networks used in computer vision to recognize objects and patterns in images. The course introduces the primary methods to identify and locate objects within an image or video. Have you ever wondered how it is possible to split an image into different sections to form ascertaining objects? This course will explore these processes and teach you the various dense sampling methods for detecting objects in an image or video. You will discover that the You Only Look Once (YOLO) architecture utilizes two fully connected layers. In contrast, the single-shot detector network uses convolutional layers of varying sizes for detecting objects. The course will reveal the significance of RetinaNet to fill the imbalances and inconsistencies of YOLO and single-shot detector methods while dealing with extreme foreground-background classes.

Next, the course highlights how the dense sampling methods give less importance to high repeatability and provide dense coverage of depicted objects. Learn how image segmentation with CNN involves feeding segments of an image as input to a CNN, which labels the pixels. Discover how CNN architecture and softmax classifiers extract distinctive face features and classify faces in the fully-connected CNN layer. The task of estimating human poses and crowd counting using deep planning and CNNs is described. This process includes evaluating the configuration of body poses from a single monocular image. You will explore how CNN employs convolution, pooling, rectified linear units and fully connected layers to extract features to obtain a crowd’s density map. The course will help you master the application of CNNs for tasks such as depth estimation, super-resolution and anomaly detection.

Finally, the method of obtaining a representation of a scene’s spatial structure, recovering the three-dimensional shape and appearance of objects in imagery using CNNs, is explained. You will be taught how to use CNNs to detect events that deviate from the standard by not following the rest of the pattern. Following this, the application of recurrent neural networks (RNNs) in analyzing computer vision problems is explained. You will explore how RNNs can work with sequences such as text, sound, videos, and finance data and how they generate captions for images. You will also explore how backpropagation through a time training algorithm is used to update weights in RNNs, like long short-term memories. Lastly, you will study how convolutional and recurrent neural networks play a vital role in producing a relevant label to a video, given its frames. This is an enlightening course that will interest those studying computer science or those interested in these topics. Why wait? Sign up today and start learning about convolutional and recurrent neural networks’ functions in processing temporal and spatial information.

Start Course Now