QA Official

VGG16

https://qaofficial.com/post/2019/04/18/36798-vgg16.html 2019-04-18
Models from the BMVC-2014 paper “Return of the Devil in the Details: Delving Deep into Convolutional Nets” The models are trained on the ILSVRC-2012 dataset. The details can be found on the project page or in the following BMVC-2014 paper: Return of the Devil in the Details: Delving Deep into Convolutional Nets K. Chatfield, K. Simonyan, A. Vedaldi, A. Zisserman British Machine Vision Conference, 2014 (arXiv ref.

ajax Submit Data to Background

https://qaofficial.com/post/2019/04/18/68562-ajax-submit-data-to-background.html 2019-04-18
ajax is a function commonly used in web development. its full name is Asynchronous JavaScript And XML. it can send data in the background. the common methods are get and post.I will take submitting forms as an example to explain the use of ajax (I use ajax encapsulated by jquery library). Html5 form code: <form id="loginForm" name="loginForm"> <table> <tr> <td><div align="right" >User Name</div></td> <td><input type="text"name="username"></td> </tr> <tr> <td><div align="right">Password</div></td> <td><input

kaggle cat vs dog

https://qaofficial.com/post/2019/04/18/36645-kaggle-cat-vs-dog.html 2019-04-18
when i studied cnn before, i took cat vs dog on kaggle to practice my hands. This model is written by pytorch and is a simple cnn with two-layer convolution.The code is not too difficult, stick it out and have a look, should be able to understand, then don't explain. import os import torch import torch.nn as nn from torch.autograd import Variable import numpy as np import cv2 import torch.

Actual Combat Machine Learning from scratch (3)-Naive Bayesian Algorithm

https://qaofficial.com/post/2019/04/18/73090-actual-combat-machine-learning-from-scratch-3-naive-bayesian-algorithm.html 2019-04-18
Naive Bayesian algorithm principle is actually relatively simple, which is based on Bayesian principle.Let's first introduce the Bayesian principle.Conditional probability: P(A|B) is the probability that a occurs under the condition that b occurs.And P(AB) = P(A|B) *P(B) = P(B|A) *P(A), from which P(B|A) = [P(A|B) *P(B)]/P(A) can be deduced. this formula gives a formula for the conversion between P(A|B) and P(B|A).This is also the mutual conversion between a priori probability

Basis of Machine Learning-Various Learning Methods (17)-Metric Learning

https://qaofficial.com/post/2019/04/18/73109-basis-of-machine-learning-various-learning-methods-17-metric-learning.html 2019-04-18
metric learning distance measurement plays a decisive role in the performance of many machine learning methods: for example, in classification methods, K-nearest neighbor classifier and stone method using Gaussian stone;Among the clustering methods, K-means clustering and spectral clustering are closely related to distance measurement.Professor Eric Xing of Carnegie Mellon University's Department of Machine Learning proposed distance measurement learning in 2003.A good distance measure can be applied to different applications according to the structure and distribution of data.

Bayesian Image Classification Experiment

https://qaofficial.com/post/2019/04/18/73084-bayesian-image-classification-experiment.html 2019-04-18
Bayesian classifier One Experimental Purpose Combine pattern recognition method with image processing technology, master the basic method of image classification by using Bayesian classifier with minimum misclassification probability, and deepen the understanding of basic concepts through experiments. 2 Experimental Principle Bayesian image classification design is based on Bayesian Decision Theory.Bayesian Decision Theory is the basic theory of statistical pattern recognition, which assumes, first, that the probability distribution of all kinds of population is known;Second, the number of categories to be classified is certain.

Big Conjecture of Artificial Intelligence Trend in the Next Ten Years

https://qaofficial.com/post/2019/04/18/72955-big-conjecture-of-artificial-intelligence-trend-in-the-next-ten-years.html 2019-04-18
Big Conjecture of Artificial Intelligence Trend in the Next Ten Years The smart products that most people come into contact with are nothing more than smart phones, smart homes, Intelligent hardware, etc. However, intelligence at this stage is just beginning. 1. outbreak of rt era With a major breakthrough in Cloud robotics's technology of learning from each other and sharing knowledge, the production cost of small household auxiliary robots has been greatly reduced, and it is expected that an emerging market of at least 41.

Common Optimization Problems and Skills of OpenCV

https://qaofficial.com/post/2019/04/18/73067-common-optimization-problems-and-skills-of-opencv.html 2019-04-18
[1] Free Mat Image Memory Space: Rect rect; rect.x = -10; rect.y = -10; rect.height = 100000; rect.width = 20000; rect &= Rect(0, 0, src.cols, src.rows);//求交集 cv::Mat crop_img = src(rect);In the example above, the size of the original image src =200*200, which needs to be trimmed to Rect = [-10,-10,10,000,20,000]. In order to avoid cutting Rect

Computer Hotspot Direction, History and Future

https://qaofficial.com/post/2019/04/18/72982-computer-hotspot-direction-history-and-future.html 2019-04-18
近几年,比较火热的方向是:自然语言处理、自动驾驶、深度学习、人工智能、大数据、虚拟化、计算机视觉 依次解释一下: 大数据: 1.背景 随着互联网和移

Depth Camera Ranging Principle Brief Introduction

https://qaofficial.com/post/2019/04/18/73020-depth-camera-ranging-principle-brief-introduction.html 2019-04-18
Deep Packet Inspection Main Technical Methods:1. Binocular matching (dual RGB cameras+optional lighting system)Triangulation principle is that the Disparity between the abscissa of the target point imaged in the left and right views is inversely proportional to the distance from the target point to the imaging plane: Z = ft/d;Get depth information.Binocular matching uses triangulation principle and is based on image processing technology. Matching points are obtained by finding the same feature points in two images, thus obtaining depth values.