Recently doing a yolov3 project, the computer is a MAC, there have been various error, then write a summary (error), use all sorts of problems yolo, continuously updated MAC
(all of the following commands are using the terminal execution)
A, install homebrew
* web site: https://brew.sh/
Terminal to perform installation instructions
* instructions below
The/usr/bin/ruby - e "$(curl - fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
Second, download wget
Command:
The brew install wget
If there is no this will not be able to download the files needed for the next yolo
Three, according to the official website
Git clone https://github.com/pjreddie/darknet
CD darknet
Make
*
If there is:
XCRUN: ERROR: INVALID ACTIVE DEVELOPER PATH (/LIBRARY/DEVELOPER/COMMANDLINETOOLS), MISSING XCRUN AT:
Execution: xcode - select - install
*
If there is:
Error: the RPC failed; The curl 56 SSLRead () return the error - 36
Execution:
1, git config - global HTTP. PostBuffer 524288000
Or
2, sudo xcode - select - switch/Applications/xcode. The app/Contents/Developer
If you only have one xcode, so this order no problem; If there are multiple xcode, needs to be modified into is as follows:
Sudo xcode - select - switch/Applications/xcode 7.3.1. The app/Contents/Developer
(center for the version number, of course, you are the best only one xcode, the other uninstall is ok)
Remember after every time need to use yolo, all need to CD darknet
After
./darknet
If the output
Usage:/darknet & lt; Function>
We can start the next
Four, the use of the training model for testing
First of all, or a CD darknet
Then
Wget https://pjreddie.com/media/files/yolov3.weights
(before downloading the wget now will be very smoothly)
Then
./darknet detect CFG/yolov3 CFG yolov3. The weights of data/t JPG
You will get this output:
0 conv 32 3 x 416 x 416 x 3-3/1 & gt; 416 x 416 x 32 0.299 BFLOPs
1 conv 3/2 3 x 416 x 416 x 32-64 & gt; 208 x 208 x 64 1.595 BFLOPs
.
105 conv 255 1 x 1/1 52 x x 256-52 & gt; 52 52 x 255 x 0.353 BFLOPs
Detection of 106
Truth_thresh: Using the default '1.000000'
Loading weights from yolov3. Weights... Done!
The data/t JPG: Predicted in 0.029329 seconds.
Dog: 99%
Truck: 93%
Bicycle: 99%
The top bar - to - personal - Mac OS driver - the user - (your user name, there is a small house logo) - darknet
This is your download darknet directory,
Folder by the name predictions. The JPG file, is the output of the matter with you,
If there is:
- bash:./darknet: is a directory
Execution:
Sudo apt - get the install g + +
(there are many kinds of this problem, I also forgot my isn't such a solution... )
If there is:
PKG - config command not found
Execution:
The brew install PKG - config
Five, opencv
Opencv is really a big pit
Below is I after several attempts to get the feasible method:
1, the brew install opencv @ 2
Now the default installation 4th edition and above, high version yolo cannot use, this is to install the second version of the
2, terminal operation will appear your installation directory, to the folder, to the installation directory of lib - pkgconfig, find opencv. PC
Then it moves to the
/usr/local/lib/pkgconfig
There are many in this folder. The PC file, if there is no even if the wrong,
Opencv. PC probability in installation position for the place where the pkgconfig signs, if not, on its own to find it, and then moving to the position of the above said,
If the installation, opencv. PC has in/usr/local/lib/pkgconfig, can directly perform the next step,
3, the following can be used to identify the video and camera real-time detection the
./darknet detector demo CFG/coco. Data CFG/yolov3 CFG yolov3. Weights & lt; Video file>