PPT-Image Classification Convolutional networks - Why

Convolutions Reduce parameters Capture shiftinvariance location of patch in image should not matter Subsampling Allows greater invariance to deformations Allows the capture of large patterns with small filters. RECOGNITION. does size matter?. Karen . Simonyan. Andrew . Zisserman. Contents. Why I Care. Introduction. Convolutional Configuration . Classification. Experiments. Conclusion. Big Picture. Why I . care. using Convolutional Neural Network and Simple Logistic Classifier. Hurieh. . Khalajzadeh. Mohammad . Mansouri. Mohammad . Teshnehlab. Table of Contents. Convolutional Neural . Networks. Proposed CNN structure for face recognition. Kuan-Chuan. Peng. Tsuhan. Chen. 1. Introduction. Breakthrough progress in object classification.. 2. O. . Russakovsky. . et al. . ImageNet. . large scale visual recognition challenge. .. . arXiv:1409.0575, 2014.. Neural . Network Architectures:. f. rom . LeNet. to ResNet. Lana Lazebnik. Figure source: A. . Karpathy. What happened to my field?. . Classification:. . ImageNet. Challenge top-5 error. Figure source: . Presenter: . Yanming. . Guo. Adviser: Dr. Michael S. Lew. Deep learning. Human. Computer. 1:4. Human . v.s. . Computer. Deep learning. Human. Computer. 1:4. Human . v.s. . Computer. Deep Learning. Why better?. Sergey Zagoruyko & Nikos Komodakis. Introduction. Comparing Patches across images is one of the most fundamental tasks in computer vision. Applications include structure from motion, wide baseline matching and building panorama. By, . . Sruthi. . Moola. Convolution. . Convolution is a common image processing technique that changes the intensities of a pixel to reflect the intensities of the surrounding pixels. A common use of convolution is to create image filters. Sergey Zagoruyko & Nikos Komodakis. Introduction. Comparing Patches across images is one of the most fundamental tasks in computer vision. Applications include structure from motion, wide baseline matching and building panorama. Last time. Linear classifiers on pixels bad, need non-linear classifiers. Multi-layer . perceptrons. . overparametrized. Reduce parameters by local connections and shift invariance => Convolution. Linear classifiers on pixels are bad. Solution 1: Better feature vectors. Solution 2: Non-linear classifiers. A pipeline for recognition. Compute image gradients. Compute SIFT descriptors. Assign to k-means centers. person 1. person 2. horse 1. horse 2. R-CNN: Regions with CNN features. Input. image. Extract region. proposals (~2k / image). Compute CNN. features. Classify regions. (linear SVM). Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation. Article and Work by. : Justin . Salamon. and Juan Pablo Bello. Presented by . : . Dhara. Rana. Overall Goal of Paper. Create a way to classify environmental sound given an audio clip. Other methods of sound classification: (1) dictionary learning and (2) wavelet filter banks . person. grass. trees. motorbike. road. Evaluation metric. Pixel classification!. Accuracy?. Heavily unbalanced. Common classes are over-emphasized. Intersection over Union. Average across classes and images. An overview and applications. Outline. Overview of Convolutional Neural Networks. The Convolution operation. A typical CNN model architecture. Properties of CNN models. Applications of CNN models. Notable CNN models.