Xử lý ảnh trong cơ điện tử: Machine Vision. Chapter 6. Image Segmentation87

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 Image Segmenta Fundamentals Point, Line, and Edge Detection Thresholding Image Segmentation Using Deep Learning Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Fundamentals ➢ Let R represent the entire spatial region occupied by an image image segmentation as a process that partitions R into n subreg R n , such that: where Q(Rk) is a logical predicate defined over the points in se set Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Fundamentals ➢ Two regions R i and R j are said to be adjacent if their union fo set ➢ The regions are said to disjoint If the set formed by the union o not connected ➢ The fundamental problem in segmentation is to partition an ima that satisfy the preceding conditions ➢ Segmentation algorithms for monochrome images generally are of two basic categories dealing with properties of intensity valu and similarity ▪ Edge-based segmentation ▪ Region-base segmentation Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Fundamentals (a) Image of a constant intensity region (b) Boundary based on intensity discont segmentation (d) Image of a texture region (e) Result of intensity discontinuity the large number of small edges) (f) Result of segmentation based on region pro Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Fundamentals ❖ Traditional Image Segmentation techniques Chapter Image Segmenta Fundamentals Point, Line, and Edge Detection Thresholding Image Segmentation Using Deep Learning Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Point, Line, and Edge Detec ➢ Detecting sharp, local changes in intensity ➢ The three types of image characteristics: isolated points, lines, ➢ Edge pixels: the intensity of an image changes abruptly ➢ Edges (or edge segments): sets of connected edge pixels ➢ Edge detectors: local image processing tools designed to detect Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Point, Line, and Edge Detec ➢ A line: ▪ a (typically) thin edge segment ▪ the intensity of the background on either side of the line is eith or much lower than the intensity of the line pixels ▪ “roof edges” ➢ Isolated point: a foreground (background) pixel surrounded by (foreground) pixels Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Point, Line, and Edge Detec ❖ Background ➢ an approximation to the first-order derivative at an arbitrary po dimensional function f(x) x = for the sample preceding x and Rafael C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) x = -1 for the sam Image Segmentation Using Deep ❖ Deep Learning-based methods ➢ Like most of the other applications, using a CNN for semantic the obvious choice ➢ When using a CNN for semantic segmentation, the output is als rather than a fixed length vector A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Convolutional neural networks for segmentation ➢ The architecture of the model contains several convolutional la activations, batch normalization, and pooling layers ➢ The initial layers learn the low-level concepts such as edges an ➢ The later level layers learn the higher level concepts such as di A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Convolutional neural networks for segmentation Spatial tensor is downsampled and converted to a vector A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Convolutional neural networks for segmentation Encoder-Decoder architecture A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Skip connections Encoder-Decoder with skip connections A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Transfer learning Transfer learning for Segmentation A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Loss function ➢ Each pixel of the output of the network is compared with the co in the ground truth segmentation image We apply standard cro on each pixel In binary classification, wher 2, cross-entropy can be calcu If M>2 (i.e multiclass classif loss for each class label per o A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation ➢ Dataset: The first step in training our segmentation model is to dataset ➢ Data augmentation: Example of image augmentati A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation ➢ Building the model: define our segmentation model with skip c ➢ Choosing the model: ▪ Choosing the base model: select an appropriate base network → 16, MobileNet, Custom CNN, … ▪ Select the segmentation architecture: FCN, SegNet, UNet, PSPN A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation Architecture of FCN32 A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation Architecture of SegNet A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation Architecture of UNet A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation Architecture of PSPNet A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation ➢ Choosing the input size: If there are a large number of objects i input size shall be larger The standard input size is somewhere to 600x600 ➢ Training: output are the model weights ➢ Testing: get predictions from a saved model A Beginner's guide to Deep Learning based Semantic Segmentation using Keras | Divam Gupta Image Segmentation Using Deep ❖ Implementation Image Segmentation results of DeepLabV3 on sample images Photo Credit: https://arxiv.org/pdf/2001.05566 pdf ... “Digital image processing,” Pearson (2018) 1 Fundamentals ❖ Traditional Image Segmentation techniques Chapter Image Segmenta Fundamentals Point, Line, and Edge Detection Thresholding Image Segmentation.. .Chapter Image Segmenta Fundamentals Point, Line, and Edge Detection Thresholding Image Segmentation Using Deep Learning Rafael C Gonzalez, Richard E Woods, “Digital image processing,”... C Gonzalez, Richard E Woods, “Digital image processing,” Pearson (2018) Chapter Image Segmenta Fundamentals Point, Line, and Edge Detection Thresholding Image Segmentation Using Deep Learning