Python Computer Vision Project Practice

This is a Bayesian model of computer Vision small project. I hope you will know how the general Computer Vision Project is operated through this simple project.

I'm going to start with the topic here. I want to be interested in the children's shoes spend a week thinking and implementation with Python. A week later I'm going to post my detailed details and code.

We hope that we can apply the knowledge of machine learning and computer vision to practice through this simple project.

Based on OpenCV to write a picture classification and classification program based on naive Bayesian

Requirements:

[1] in Google's image search engine input "flower" and "airplane", respectively download m (>100) Zhang "flower" Picture and N (>100) "Sky" picture, as a positive and negative sample of the data set; Note: These two types of images can also be arbitrarily selected, for example "chair" and "car"

[2] The visual lexicon is constructed using the following methods: Randomly dividing the picture blocks of P (p>500) 25x25 from the pictures in each training set, calculating the color histogram feature (6x6x6 interval) of the image block, and then adding the blocks of these images (collectively owning p* (m+ N) Use Kmeans to form K (k>1000) classes as dictionaries; Note: Kmeans can call OpenCV directly or use other toolkits to cluster.

[3] Training or testing, to the test of the image of dense sampling, with histogram intersection distance (calculated formula). The number of times each word is calculated using a recent neighbor method.

[4] Using 50 percent cross-validation to calculate the average precision and recall;

datasets You can consider writing a crawler to Google to download images ( Let's say you can teach me how to operate a reptile, huh? if not, can I ask for data set QQ #14 # 14# 626.

Python Computer Vision Project Practice