In that case, it shows many pages of errors in the C opencv headers (even for an empty program that simply includes an opencv header and links to an opencv lib) - so it does find the headers and the libraries. On a clean ubuntu 18.04 results in an opencv install that works if you use C++ opencv functions, but does not work if you use C opencv functions. Other things you could look at: looks like sudo apt-get install libopencv-dev If you do not want to download the source and compile, If you download the source and compile it as shown above, you are done, you can stop here, this worked for me on xubuntu 20.04 Why ninja instead of make? ninja can build in parallel on multiple processors so it is faster Īnd now you can compile your source that worked in 16.04 with opencv 2.4.9 Sudo apt install ninja-build ant cmake build-essentialĪt this point you can now compile with opencv2.4! however, at runtime, you may run into an "error while loading shared libraries" which is fixed by this additional command: sudo /sbin/ldconfig -v If you don't need any special exotic options, the basic build and install instructions can be summarized like this:Ĭlick on green button and download opencv2.4 zip release from the above github link then at the commandline: (note that the build directory must be on a linux filesystem drive such as ext4) unzip opencv-2.4.zip Instructions to build opencv from source after downloading the 2.4 branch as a zip and unzipping it can be found here: You can get the 2.4 branch source here and build that, I think this is the best solution, and it worked for me on ubuntu server 18.04 My steps were the following: git clone -recursive In my case, instead of using the opencv/opencv repository, I used branch 2.4 from as suggested in this post (it uses 2.4.13 though). Jmarina's answer helped me a lot with this issue.
0 kommentar(er)
