Lack of documentation
Hi,
OpenCV is a large useful and complicated library and it is growing fast in size. However, many of the modules/functions which are being added are difficult/nearly impossible to use or understand due to the lack of basic documentation. I am talking about basic documentation and not exhaustive one. For instance, the stitching module, which is a large chunk of code, but very little documentation. What does the member variable indices_ in the stitcher class used for ? Perhaps it would be best to add more documentation regulations to contributed packages: 1. Reference to a relevant article/s for each algorithm. 2. For each variable use meaningful names ( variable "k" in a loop is meaningless). 3. Add a brief description of each function above that function, and a word or two near each member variable, and local variables. 4. Whenever it can help - add comments in the code.
This will make the OpenCV code much more readable and usable. After all, from what I understand the hard work put into OpenCV is meant to assist as many users as possible, and not to create an exclusive club of few people who actually understand what's going inside.... so I hope....
If we would limit people in how they have to format their internal code when submitting new stuff, then the amount of submissions would drop even further. This is a open source software library with people using their spare time to devellop stuff, besides the basic staff that is behind the library. If you use the functionality provided, this is documented, even the papers used are documented, as seen here. Once you want to dive into code, you are on your own, using papers and your knowledge to analyse the provided code. I think the C++ API however is as clear as it can be, and docs are growing each day!