Deep Learning for Computer Vision Latest

DeepLearning.AI Convolutional Neural Networks Advanced Learning Specialization Course (Review)

Convolutional Neural Networks - Computer vision




Google plus

Andrew Ng is legendary for his Stanford machine learning course at Coursera.

In 2017, he additionally revealed five elements of the course in in-depth learning at Coursera, entitled "Deep Specialization in Learning". Pc Convolutional Neural Networks.

This course offers a superb introduction to a deep learning technique in pc imaginative and prescient purposes for practitioners who already know the fundamentals of in-depth studying. It doesn't focus an excessive amount of on math and doesn't include any code. As an alternative, it is designed to develop intuitions for key applied sciences used in the art.

On this submit, you will study concerning the course of convolutional networks taught by Andrew Ng from profound studying to a pc. [19659003] After studying this message, you understand:

  • The course is a broader course of in-depth studying offered by
  • The course is just not free and requires registration and enrollment at Coursera, regardless that all videos are available totally free on YouTube.
  • The course offers a superb introduction to deep studying for computer-aided builders who know the fundamentals of in-depth learning.


Overview [19659012] This tutorial is divided into 5 sections; they’re:

  1. Overview of Specialization in Deep Learning
  2. Change of Convolutional Neural Networks
  3. Course Videos on YouTube
  4. Discussion and Evaluate

Deep Learning Specialization Overview

Andrew Ng is a Machine Scholar Recognized for Learning the Stanford Machine Learning Course publicly and later for basic practitioners and Coursera

He’s also the founder of Coursera and former CEO of Google Brain and Baidu.

In 2017, he began a brand new website referred to as, which offers in-depth primary coaching for common practitioners (eg developers) on programs out there by means of his Coursera platform (which requires an order).

Your complete course is divided into five sub-courses; they’re:

  • Course 1: Neural Networks and Deep Learning
  • Course 2: Enhancing Deep Neural Networks
  • Course three: Constructing Engineering Language Tasks
  • Course Four: Convolutional Neural Networks (Our Priority)
  • Course 5: Sequence Models [19659009] The courses are in the form of video tutorials and Andrew presents in the identical practical type as his famous Machine Learning course.

    The course is designed for builders who are novices

    Convolutional Nervous Networks Breakdown

    Word is a profound studying for a computer vision course referred to as "Convolutional Neural Networks."

    Convolutional Neural Networks – Pc Imaginative and prescient

    This course is designed to teach developers how convolutional neural networks work and how they’re used for widespread pc vision duties.

    This course teaches you easy methods to construct convolutional neural networks and apply it to image knowledge. Because of in-depth studying, your pc's imaginative and prescient works a lot better than just two years in the past, and this lets you use a wide range of fascinating purposes from protected, unbiased driving, to express face recognition, to automated studying of radiological pictures.

    – About this course

    It is divided into four weeks; they’re:

    • Week 1: Foundations of Convolutional Neural Networks
    • Week 2: Deep Convolution Fashions: Case Studies
    • Week 3: Object Detection
    • Week Four: Particular Packages: Face Detection and Nervous Transplant [19659010] Every Week divided into 10 -12-themed subjects, every masking a short video lasting a couple of minutes and as much as 15 minutes.

      A lot of the subjects are introduced in practically very little math. What little mathematics is concerned has targeted on subjects resembling calculating loss features or calculating the number of parameters (weights) in the mannequin.

      Along with the shortage of arithmetic, the course additionally doesn’t present code but focuses on serving to the viewer develop instinct on the methods being handled

      Week 1. The primary week is meant to familiarize itself with an important sort of neural network used for pc vision issues: convolutional neural community or CNN. Subjects give attention to how convolution layers work, filters, cushions, steps, and the associated layer layer.

      Week 2. The second week focuses on essential milestones for the creation of powerful CNN fashions, akin to LeNet, and models developed for SuiteNet corresponding to AlexNet, ResNet and Inception. When every milestone has been discussed, innovation in design is described in a means that explains why the model is efficient and the way know-how can be used extra usually. The week ends with knowledge preparation, together with knowledge insertion and transfer studying.

      Week three. The third week focuses on object detection, which introduces easier image classification, image localization, and landmark detection issues. These are in-depth pc imaginative and prescient studying purposes and essential sub-technologies of the YOLO technique are introduced step by step and built into an entire work system.

      Week Four. Lastly, the last week ends with face recognition and recognition that develops the methods required by such a system, together with one shot learning, siamese nets and an appropriate loss motion. The week is divided into two elements, and the opposite half focuses on neurotransmission, but know-how that is purely aesthetic is fun.

      Course Movies on YouTube

      The entire division in fact subjects

      • Week 1: Foundations of Convolutional Neural Networks
        • Pc Imaginative and prescient
        • Edge Detection Instance
        • Extra Edge Detection
        • Padding
        • Excursions
        • Convolutions Over Volume
        • One Spherical of Convolution Network
        • Simple Example of Convolution Network
        • CNN Example
        • Why Convolutions?
        • Yann LeCun Interview
      • Week 2: Deep Convolution Fashions: Case Research
        • Why Evaluate Case Research?
        • Classical Networks
        • ResNets
        • Why ResNets Work
        • Networks in Networks and 1 × 1 convolutions
        • Preliminary Community Motivation
        • Getting Started
        • Open Source Implementation
        • Open Source Implementation
        • Transfer Learning
        • Knowledge Augmentation
        • Pc imaginative and prescient state
      • Week 3: Object Detection
        • Object Localization
        • Landmark Identification
        • Object Identification
        • Implementing Sliding Home windows Conversions
        • Cropping Box Predictions
        • Connections within the Union
        • Non-max Damping
        • Anchor Instances
        • algorithm
        • (elective) Space suggestion
      • Week Four: Special Purposes: Face Identification and Nervous Transplant
        • What is Face Detection?
        • One Shot Learning
        • Siamese Network
        • Triplet Loss
        • Face Verification and Binary Classification
        • What is Nervous Transfers?
        • What’s ConvNets Learning Deep?
        • Publishing Function
        • Content material Value Function
        • Fashion Costing
        • 1D and 3D Generalizations

      Course videos are also obtainable by way of YouTube.

        Convolutional Neural Networks taught by Andrew Ng

      Convolutional Neural Networks taught by Andrew Ng

      YouTube Playlist The course can also be out there, although some week 3 movies are disabled:

      Videos have C4WnLnn naming coverage, the place Wn refers back to the week number (1-Four) and Lnn refers back to the lecture quantity (eg 01-12). Pay attention to the order of the videos at week 3;

      Two movies lacking from playlist; they’re:

      Notice that week 1 is a video interview with Yann LeCun, the inventor of Convolutional neural networks. That is part of a collection of movies entitled "Heroes of Challengers for Deep Learning", which can also be obtainable on YouTube.

      You’ll be able to simply watch all your movies in a number of hours.

      Dialogue and Evaluation

      I looked at all

      It's an amazing course, and I feel it does a superb job of creating intuitions on handled subjects, including CNNs, milestones, object identification, face recognition, and elegance transfer

      Introduction to convolution layers and related subjects, associated to the cushioning, step and so on., is probably one of the cleanest performances I have seen. I additionally discovered a milestone for week 2 and week three on CNN models and object detection very clearly. Week 4 was high quality, properly started, however acquired very messy once we took part within the transmission of nerve cells.

      In case you are already acquainted with the fundamentals of CNN, like many in-depth learning practitioners, I recommend focusing on Week 3 and Week Four. Week three is particularly nice if you dive into object-killer software parts and difficult YOLO technique to handle it.

      YOLO is a vital matter that must be addressed, but it might have been simplified if it was targeted on an easier (easier to implement technique), corresponding to quick or quicker R-CNN.

      I discovered myself watching virtually each minute of the course by skipping just some messy mathematical descriptions. In truth, joining the reason of loss features was maybe the least fascinating and perhaps non-compulsory part of the courses.

      I've never found a video as an efficient software for educating math. It have to be taken slowly, with clear (LaTeX) equations, rationalization and code

      In case you are new to deep studying as a computer vision, but know the fundamentals of in-depth learning, I recommend courses, especially watching movies.

      Read extra

      This section incorporates extra assets on the topic if you wish to go deeper.


      In this submit, you discovered a breakdown of the analysis of the convolutional neural community taught by Andrew Ng for a deep studying of pc vision

      Particularly, you study:

      • The course is a sub-course of deeper learning from in-depth learning
      • The course isn’t free , and it requires subscription and registration for Coursera, although all movies can be found without spending a dime on YouTube.
      • This course offers a superb introduction to in-depth learning for computer-aided builders who realize it. the basics of in-depth learning

      Do you’ve got any questions?
      Ask your question within the comments under and do my greatest



      Google plus