Deep Learning For Computer Vision Research Based PDF Notes

Free Download Deep Learning for Computer Vision Research Based PDF Notes. This notes is very useful and helpful for developer, researcher and students. In this notes you’ll learn how to solve computer vision problems by deep learning method. If you are interested in deep learning or machine learning you will easily understand its concepts.

In this notes you’ll learn how to solve real-word problem by using of deep learning and computer vision. This practical guide teach you to build your knowledge and you’ll learn how to solve image classification and face recognition problems. This notes for everyone who interested in deep learning or machine learning and wants to gain more knowledge in computer vision.


Tutorial Deep Learning for Computer vision Research Based PDF Notes
Format  PDF



 Data Augmentation

  •  What Is Data Augmentation?
  • Visualizing Data Augmentation
  •  Comparing Training With and Without Data Augmentation
  •  The Flowers-17 Dataset
  •  Aspect-aware Preprocessing
  •  Flowers-17: No Data Augmentation
  •  Flowers-17: With Data Augmentation
  • Summary

 Networks as Feature Extractors

  • Extracting Features with a Pre-trained CNN
  •  What Is HDF5?
  •  Writing Features to an HDF5 Dataset
  •  The Feature Extraction Process 37
  •  Extracting Features From Animals
  • Extracting Features From CALTECH-101
  •  Extracting Features From Flowers-17
  •  Training a Classifier on Extracted Features
  •  Results on Animals
  •  Results on CALTECH-101
  •  Results on Flowers
  • Summary

Understanding rank-1 & rank-5 Accuracies

  •  Ranked Accuracy
  •  Measuring rank-1 and rank-5 Accuracies
  • Implementing Ranked Accuracy
  •  Ranked Accuracy on Flowers-17
  •  Ranked Accuracy on CALTECH-101
  • Summary

Fine-tuning Networks

  • Transfer Learning and Fine-tuning
  • Indexes and Layers
  • Network Surgery
  •  Fine-tuning, from Start to Finish
  • Summary

Improving Accuracy with Network Ensembles

  • Ensemble Methods
  • Jensen’s Inequality
  •  Constructing an Ensemble of CNNs
  •  Evaluating an Ensemble
  • Summary

Advanced Optimization Methods 

  • Adaptive Learning Rate Methods
  • Adagrad
  • Adadelta
  • RMSprop
  • Adam
  • Nadam
  • Choosing an Optimization Method
  •  Three Methods You Should Learn how to Drive: SGD, Adam, and RMSprop
  • Summary

Optimal Pathway to Apply Deep Learning

  •  A Recipe for Training
  • Transfer Learning or Train from Scratch
  •  Summary

Working with HDF5 and Large Datasets 

  •  Downloading Kaggle: Dogs vs. Cats
  • Creating a Configuration File
  • Your First Configuration File
  • Bulding the Dataset
  • Summary

Competing in Kaggle: Dogs vs. Cats 

  • Additional Image Preprocessors
  •  Mean Preprocessing
  •  Patch Preprocessing
  • Crop Preprocessing
  •  HDF5 Dataset Generators
  • Implementing AlexNet
  •  Training AlexNet on Kaggle: Dogs vs. Cats
  •  Evaluating AlexNet
  • Obtaining a Top-5 Spot on the Kaggle Leaderboard
  •  Extracting Features Using ResNet
  •  Training a Logistic Regression Classifier
  • Summary


  • The Inception Module (and its Variants)
  •  Inception
  • 11.1.2 Miniception
  •  MiniGoogLeNet on CIFAR-10
  •  Implementing MiniGoogLeNet
  •  Training and Evaluating MiniGoogLeNet on CIFAR-10
  •  MiniGoogLeNet: Experiment #1
  • MiniGoogLeNet: Experiment #2
  •  MiniGoogLeNet: Experiment #3
  •  The Tiny ImageNet Challenge
  •  Downloading Tiny ImageNet
  •  The Tiny ImageNet Directory Structure
  •  Building the Tiny ImageNet Dataset
  •  DeeperGoogLeNet on Tiny ImageNet
  •  Implementing DeeperGoogLeNet
  •  Training DeeperGoogLeNet on Tiny ImageNet
  •  Creating the Training Script
  •  Creating the Evaluation Script
  •  DeeperGoogLeNet Experiments
  • Summary


  • ResNet and the Residual Module
  • Going Deeper: Residual Modules and Bottlenecks
  •  Rethinking the Residual Module
  •  Implementing ResNet
  • ResNet on CIFAR-10
  •  Training ResNet on CIFAR-10 With the ctrl + c Method
  •  ResNet on CIFAR-10: Experiment #2
  • Training ResNet on CIFAR-10 with Learning Rate Decay
  •  ResNet on Tiny ImageNet
  •  Updating the ResNet Architecture
  •  Training ResNet on Tiny ImageNet With the ctrl + c Method
  •  Training ResNet on Tiny ImageNet with Learning Rate Decay
  • Summary





About the author


Leave a Comment