Introduction
This blog is going to be used to follow up the paper I read about sensor topics. Therefore, I can go back and check easily if I need to review or remember some topics. In master degree, I worked on image classification , image segmentation, and some basical Deep Learning knowledges. The papers collected are read after that. Most of them are support and rich my knowledge system during working and studying.
Image Processing and Preparing
Dataset Information
For the projects, I am searching for more informtion about dataset used for Machine Vision topic. In master degree, I was using the dataset directly for researching. However, at this point I have to build a dataset for the special task. Therefore, I have to research how to create the dataset can be used. The project is similar to diagnoisis cancer from medical images, so I began from there.
“Cancer Diagnosis in histopathological image: CNN based approach”
Datasets for breast cancer: Breast Cancer for breast (WDBC) cancer Wisconsin Original Data Set (UC Irvine Machine Learning Repository), MITOS- ATYPIA-14, and BreakHis.
This used the labeled (benign/malignant) input image form the raw pixels and highlighted the visual patterns, and then utilize those patterns to distinguish between the visual patterns, and then utilize those patterns to distinguish betwwen non-cancerous and cancer containing tissue, working akin to digital staining, which spotlights image segments crucial for diagnostic decisions, with the help of a classifier network.
Most of the pixels in the image are redundant and do not contribute substantially to the intrinsic information of an image. While dealing with AI networks, it is required to eliminate them to avoid unnecessary computational overhead. This can be achieved by compression techniques. We begin the implementation of our deep net by processing the images in the dataset.
Feature learning is a crucial step in the classification process for both human and machine algorithm. A study has shown that the human brain is sensitive to shapes, while computers are more sensitive to patterns and texture.
“A Dataset for Breast Cancer Histopathological Image Classification”
For the project, dataset will use the synthetic data. A study has shown that the human brain is sensitive to shapes, while computers are more sensitive to patterns and texture. I am goint to check how it might will affect the training for our project.
“ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness”
Image Label
Computer Vision
Object Detection
Object Recognization
If you have any questions, please contact with tianluwu@gmail.com