MMM-FacialRecognition Training tool
I cloned the Facial Recognition training tool to my linux laptop, to generate a Training.xml file that i could put on my pi later, but its not working…
First i tried capturing with my webcam, and i see the webcam light go on, but it fails like so:
sudo python capture.py Remember to set the name list environment variable FACE_USERS Using default FACE_ALGORITM: 1 What do you want to do?  Capture training images from webcam  Convert '*.jpg' pictures from other cameras to training images --> 1 Enter the name of the person you want to capture or convert images for. --> Erik Images will be placed in ./training_data/Erik -------------------- Starting process... Capturing positive training images. Press enter to capture an image. Press Ctrl-C to quit. Traceback (most recent call last): File "capture.py", line 38, in capture.capture(preview) File "/root/MMM-Facial-Recognition-Tools/lib/capture.py", line 54, in capture input() File "", line 0 ^ SyntaxError: unexpected EOF while parsing
Then I tried just using jpg files, got no errors but no images are actually being put in the [Person’s name] folder under training data…
root@atreyu:~/MMM-Facial-Recognition-Tools# python capture.py Remember to set the name list environment variable FACE_USERS Using default FACE_ALGORITM: 1 What do you want to do?  Capture training images from webcam  Convert '*.jpg' pictures from other cameras to training images --> 2 Enter the name of the person you want to capture or convert images for. --> Erik Images will be placed in ./training_data/Erik Please enter path to images or drag and drop folder into terminal --> '/root/Documents/erik1.jpg' -------------------- Starting process... root@atreyu:~/MMM-Facial-Recognition-Tools#
I put gave it like 5 images for Erik and one for Essy, but
When i try running the training.py it looks like there are no images: i get this:
root@atreyu:~/MMM-Facial-Recognition-Tools# python train.py Remember to set the name list environment variable FACE_USERS Using default FACE_ALGORITM: 1 Which algorithm do you want to use?  LBPHF (recommended)  Fisherfaces  Eigenfaces --> 1 Reading training images... -------------------- Assign label 1 to Essy Assign label 2 to Erik -------------------- ('Read', 0, 'positive images and', 400, 'negative images.') -------------------- Training model type 1 with threshold 80 ('Training data saved to', 'training.xml') Please add or update (if you added new people not just new images) ['Essy', 'Erik'] inside config.js (mirror module) or config.py (model tester). You can change the names to whatever you want, just keep the same order and you'll be fine. Please add 1 as your choosen algorithm inside config.js (mirror module) or config.py (model tester). root@atreyu:~/MMM-Facial-Recognition-Tools#
Am i missing something?
I tried running the Facial recognition test, and it opens up the preview and detects my face, eyes, nose etc… but always says no match
Ok I figured out that it wasn’t recognizing images because I was suppose to give a folder not a file… that said, I now have it trained for three people, [Erik, Essy, Paul] but when it recognizes any of them it just says that they are Essy when I run the tester script. I updated the config.py and everything… I know it recognizes all three people because others show up as None…
Anyone have this problem
Seems to be a problem in the config.py it was setting label to 0 no matter what
Jan last edited by
Hi. I face the same issue as you, Erik. I can’t make captures via camera (same error as you wrote), and if i train my own jpgs, i can not run the face recognation without issues. Did you solve the problem ?
apm1007 last edited by
got the same problem…
gismo2006 last edited by
I hopüe I understand your problem and maybe I can help you. I use this modul since about one year. At first I had many problems but now it works perfect.
What I do: I use the trainer on my pi. Because of my experince whit an other modul, i rember its better not to use other sources (like your laptop…)
I use a usb cam not the pi cam. Because of that the traine wouldn´t work whit option one (take pictures from your webcam) So I have to choose option to (use jpg pictures)
for this i use fswebcam whit the option --no-Banner. So i take about 50 pictures of myself. Then I copied all pictures in one Folder, Pictures for example. Choose in the trainer option 2 use your own jpg pictures and paste the path to your folder. I think now or before you choose your path the trainer ask you for a name. Lets work it… After this you have a new folder in your folder facetrainer. now you can start the train.py and everthing should Ok.
reaper81 last edited by
i also have a problem withe Facial Recognition Tool. When i start “python capture.py” i get this error.
pi@raspberrypi:~/MagicMirror/modules/MMM-Facial-Recognition-Tools $ sudo python capture.py
Traceback (most recent call last):
File “capture.py”, line 18, in
import lib.capture as capture
File “/home/pi/MagicMirror/modules/MMM-Facial-Recognition-Tools/lib/capture.py”, line 12, in
from builtins import input
ImportError: No module named builtins
I deleted and cloned the module twice. The installation of the OpenCV dependencies where successfull. The installation of the python dependencies (“pip install -r requirements.txt”) failed, because there where no requirements.txt.
I hope you can help me. Thanks.
reaper81 last edited by
OK, I found a sollution by my self. The Problem was realy the python dependency. It seems in the missing “requirements.txt” where the python future package listed so I installed it by my self with
sudo pip install future --upgrade
I hope paviro fix the missing file in his project, because I like it very much.
nuckinfutz last edited by nuckinfutz
kengberg last edited by
This is aweome - I needed that fix as well - one step closer, thanx!