Skip to content
100644 40 lines (32 sloc) 1.74 KB
1884333 @liuliu modified validator script
authored Jun 27, 2012
1 Intro
2 =====
4 Around 2010, when Lian and I was working on our gesture recognition demo, out
5 of the frustration to abstract redundant image preprocessing operations into a
6 set of clean and concise functions, I started to consider moving away from the
7 stack. Why? Well, after two years, ccv is the very answer.
9 Cached Image Preprocessing
10 --------------------------
12 Many computer vision tasks nowadays consist of quite a few preprocessing
13 layers: image pyramid generation, color space conversion etc. These potentially
14 redundant operations cannot be easily eliminated within a mature API. ccv
15 provides a built-in cache mechanism that, while maintains a clean function
16 interface, effectively does transparent cache for you.
18 Easy to Embed
19 -------------
21 While it depends on quite a few libraries for the best performance and
22 complete feature, ccv's majority functionalities will still work without these
23 libraries. You can even drop the ccv source code into your project, and it will
24 work!
26 Modern Computer Vision Algorithms
27 ---------------------------------
29 One core concept of ccv development is "application driven". As a result, ccv
30 end up implementing a handful state-of-art algorithms. It includes
31 a very fast detection algorithm for rigid object (face etc.), an accurate
32 object detection algorithm for somewhat difficult object (pedestrian, cat etc.),
33 a state-of-art text detection algorithm</a>, and the long-standing feature point
34 detection algorithm.
36 For computer vision community, there is no shortage of good algorithms, good
37 implementation is what it lacks of. After years, we stuck in between either the
38 high-performance, battle-tested but old algorithm implementations, or the new,
39 shining but Matlab algorithms. ccv is my take on this problem, hope you enjoy
40 it.
Something went wrong with that request. Please try again.