Xử lý ảnh trong cơ điện tử: Machine Vision. Chapter 7. Object Recognition86

TRƯỜNG ĐẠI HỌC BÁCH KHOA XỬ LÝ ẢNH TRONG CƠ ĐIỆN Machine Vision Giảng viên: TS Nguyễn Thành Hùn Đơn vị: Bộ môn Cơ điện tử, Viện Cơ Hà Nội, 2021 Chapter Object Recogn ❖1 Introduction ❖2 Pattern Matching ❖3 Feature-based Methods ❖4 Artificial Neural Networks Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ Object recognition: localize and to classify objects ▪ General concept: ➢ training datasets containing images with known and labelled ➢ extracts different types of information (colours, edges, geo chosen algorithm ➢ for any new image the same information is gathered and dataset to find the most suitable classification Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ Applications: ➢ robots in industrial environments, ➢ face or handwriting recognition ➢ autonomous systems such as modern cars which use objec detection, emergency brake assistant and so on ➢… Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies ➢ Appearance-based method ➢ Feature-based method ➢ Interpretation Tree ➢ Pattern Matching ➢ Artificial Neural Networks Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies: Appearance-based m ➢ Face or handwriting recognition ➢ Reference training images ➢ This dataset is compressed to obtain a lower dimension subspac ➢ Parts of the new input images are projected on the eigenspace examined Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies: Feature-based Meth ➢ Characteristic for each object ➢ Colours, contour lines, geometric forms or edges ➢ The basic concept of feature-based object recognition strategies • Every input image is searched for a specific type of feature, • This feature is then compared to a database containing model verify if there are recognised objects Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies: Feature-based meth ➢ Features and their descriptors can be either found considerin feature) or after observing just small parts of the image (local fe ➢ An histogram of the pixel intensity or colour are simple exampl ➢ It is not always reasonable to compare the whole image, as illumination, position (occlusion) or rotation lead to significan recognition is not possible anymore Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies: Feature-based meth ➢ Descriptors of local features are more robust against thes algorithms with local features often outperform global feature-b Two patches of different images are cut and compared if the error between the patches is below a certain threshold Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies: Interpretation Tree ➢ Interpretation tree is a depth first search algorithm for model m ➢ Algorithms based on this approach often try to recognise objects, therefore a database containing models with known fea ➢ The feature set might consist of distance, angle and direction on the surface of the objects Simon Achatz, State of the art of object recognition techniques, Technische Universitat M Artificial Neural Netwo ➢ The Pooling Layer Max-pooling in 3D image, which is the one that we norm Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ➢ The Pooling Layer Applying Max/Sum pooling to image that applied ReLU Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ▪ Layers used to build ConvNets ➢ Fully Connected Layer (FC) Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ➢ Fully Connected Layer (FC) Fully connected layer Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ➢ Fully Connected Layer (FC) ▪ Softmax function: takes a vector of arbitrary real-valued score it to a vector of values between zero and one that sum to one Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ LeNet Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ AlexNet Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ ZFNet Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ Inception-v4 Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ VGGNet Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ VGGNet Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ ResNet Stanford University CS231n: Convolutional Neural Networks for Visual Recognition, 2020 Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ Example: LeNet Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ Example: LeNet Artificial Neural Netwo ❖ CNN - Convolutional Neural Network ➢ Example: LeNet .. .Chapter Object Recogn ❖1 Introduction ❖2 Pattern Matching ❖3 Feature-based Methods ❖4 Artificial Neural Networks Simon Achatz, State of the art of object recognition techniques,... State of the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies: Feature-based Meth ➢ Characteristic for each object ➢ Colours,... the art of object recognition techniques, Technische Universitat M Introduction ▪ General Object Recognition Strategies: Interpretation Tree Procedur Simon Achatz, State of the art of object recognition