MMM-Face-Reco-DNN

[sdetweil]

@Parvez what happens when u run the script?

Facial Recognition Embeddings
After you set up your dataset, we need to go to the embeddings for the recognitions. I prepared a script for that. It took a while to run on my Macbook and will take much longer on a Raspberry Pi, so if possible, use it on a computer with more power and then transfer the resulting file.

You will find a script called encode.py in folder tools of the module. The following arguments are default:

python3 encode.py -i ../dataset/ -e encodings.pickle -d hog

linux us case sensitive… so Parvez is not the same as parvez

[Parvez]

if any one make a vedio for me is very helpful . i will submit this project my university

[Parvez]

@sdetweil said in MMM-Face-Reco-DNN:

python3 encode.py -i …/dataset/ -e encodings.pickle -d hog

how to run this script

[sdetweil]

@Parvez

you will find a script called encode.py in folder tools of the module

so,

cd tools

[Parvez]

[sdetweil]

@Parvez looks like it is trying to detect text in the images… don’t know why

https://raspberrypi.stackexchange.com/questions/89231/tesseract-ocr-4-x-beta-for-raspberry-pi

[sdetweil]

@Parvez dist-packages/cv2 seems pretty old… cv3 is current, cv4 is also out…

[Parvez]

@sdetweil ok i will download open cv3

[Parvez]

i can’t solve it

[sdetweil]

@Parvez sorry, no idea…

[Parvez]

@sdetweil Thank you for your help…

[jimbydude]

@Parvez hi,

I took a fork of SMAI that uses recoDNN an doesn’t require any external python running (recoDNN does that), i also detail training.

https://github.com/jimbydude/MMM-AI-Face-Login

Cheers

[Seregeth]

Hi,
i have done all to import my fotos. i worked all along without any error.
is it possible to test the importet Foto set?
As i set up the Module, i will not recognice me. Is it possible to test anyway?

[sdetweil]

@Seregeth I do not know. this would have to be provided by the OCV (Open computer vision)
code, not ours

[Seregeth]

@sdetweil
Thank you very much for the feedback.

I almost thought so. But maybe someone knows, so I can narrow down where my mistake is.

If it’s because the person isn’t recognised, it’s different than if I’ve set a parameter incorrectly.

[sdetweil]

@Seregeth the module hasn’t been supported for a number of years, and I don’t remember any debugging assistance provided

[Seregeth]

This post is deleted!

[sdetweil]

@Seregeth I think you posted this to the wrong topic

has nothing to do with face reco

[Seregeth]

I have now managed to get the module to recognise me. Following this nice success, I now wanted to add other people to it.

But that is no longer possible, the process always stops:

I start:
python3 encode.py -i …/dataset/ -e encodings.pickle -d hog

then comes:
[INFO] quantifyng faces…
[INFO] processing image 1/17 - …/dataset/Pauline/img01.jpg
Killed

I then tried deleting the first image (img01.jpg) once and trying again. It worked better then, but it still broke. This time, image img01.jpg came first, then image img04.jpg, then Killed.

What could be the reason that the process is no longer executed?

[sdetweil]

@Seregeth the images are not the right quality or size?