This alwaysAI app does facial detection and recognition. It uses this python library which is based on the dlib library.
Clone this repo into a local directory. Then cd into the facial_recognition folder and run aai app configure and make the following selections:
- When prompted to choose a project, use the down arrow and select
Create new project, choosing any name you like. - When prompted to create
alwaysai.project.json, selectYes
The app.py and alwaysai.app.json files should be automatically detected and you should not need to create them.
You can find details on working with projects here.
Before running the app, you need to prepare the dataset of the people you want to recogise. For the purpose of simplicity, I am using a kaggle-dataset. Just copy the folders from the train directory into the dataset folder.
You could also create your own set by collecting around 8-10 images of a person and copy into a folder with the name into the dataset folder.
Copy the images from the val directory into the images directory for testing.You could also download images from the internet and put into the images folder.Next, cd into the scripts directory and run make. It would generate encodings.pickle which is the database of the known people.
% make
docker build --build-arg UID=502 --build-arg GID=20 -t facial-recognition-dataset-creator .
Sending build context to Docker daemon 6.144kB
Step 1/12 : FROM jjanzic/docker-python3-opencv
---> 7b19919b069b
Step 2/12 : RUN pip install dlib face_recognition imutils
---> Using cache
---> 306086a11ea4
Step 3/12 : WORKDIR /srv
---> Using cache
---> 91ffd65e9c11
Step 4/12 : ARG UNAME=ocvuser
---> Using cache
---> 66e791a6dfa7
Step 5/12 : ARG UID=1000
---> Using cache
---> e606df4aee61
Step 6/12 : ARG GID=1000
---> Using cache
---> 0c7f193b08cb
Step 7/12 : RUN groupadd -g $GID -o $UNAME
---> Using cache
---> da6c192b567a
Step 8/12 : RUN useradd -m -u $UID -g $GID -o -s /bin/sh $UNAME
---> Using cache
---> 78ca1ae1b8cd
Step 9/12 : ENV method=hog
---> Using cache
---> c8976d20ad46
Step 10/12 : USER $UNAME
---> Using cache
---> 0334c3004f05
Step 11/12 : ADD *.py *.xml /srv/
---> abc62ebefa91
Step 12/12 : CMD [ "python3", "/srv/encode.py", "-i", "/srv/dataset", "-e", "/srv/model/encodings.pickle", "-d", "${method}" ]
---> Running in fd29130e7f37
Removing intermediate container fd29130e7f37
---> cd9ab7d99083
Successfully built cd9ab7d99083
Successfully tagged facial-recognition-dataset-creator:latest
docker run -it -v /Users/abhijeet.bhatikar/AAI/Project/aai_facialrecognition/facial_recognition/scripts/../dataset:/srv/dataset -v /Users/abhijeet.bhatikar/AAI/Project/aai_facialrecognition/facial_recognition/scripts/../:/srv/model -e method=hog facial-recognition-dataset-creator:latest
[INFO] quantifying faces...
[INFO] processing image 1/105 - /srv/dataset/Amitabh/images (2).jpeg
[INFO] processing image 2/105 - /srv/dataset/Amitabh/download.jpeg
[INFO] processing image 3/105 - /srv/dataset/Amitabh/images (4).jpeg
...
...
...
...
...
...
[INFO] processing image 96/105 - /srv/dataset/jerry_seinfeld/httpgraphicsnytimescomimagessectionmoviesfilmographyWireImagejpg.jpg
[INFO] processing image 97/105 - /srv/dataset/jerry_seinfeld/httpwwweonlinecomresizewwweonlinecomeolimagesEntireSitersxjerryseinfeldjulialouisdreyfusjwjpg.jpg
[INFO] processing image 98/105 - /srv/dataset/jerry_seinfeld/httpstatictherichestimagescomcdncwpcontentuploadsJerrySeinfeldjpg.jpg
[INFO] processing image 99/105 - /srv/dataset/jerry_seinfeld/httpwwwwatchidcomsitesdefaultfilesuploadsightingBreitlingwatchJerrySeinfeldjpg.jpg
[INFO] processing image 100/105 - /srv/dataset/jerry_seinfeld/httpimagescontactmusiccomnewsimagesjerryseinfeldjpg.jpg
[INFO] processing image 101/105 - /srv/dataset/jerry_seinfeld/httpimagesrottentomatoescomimagesspotlightsnewsjerryseinfeldjpg.jpg
[INFO] processing image 102/105 - /srv/dataset/jerry_seinfeld/httpsrobertaccetturacomwpcontentuploadsjerryseinfeldheadshotjpg.jpg
[INFO] processing image 103/105 - /srv/dataset/jerry_seinfeld/httppfodcompchannelslegacyprofilejerryseinfeldpodcastjpg.jpg
[INFO] processing image 104/105 - /srv/dataset/jerry_seinfeld/httpikinjaimgcomgawkermediaimageuploadsWmIuhdsrcedidjpgjpg.jpg
[INFO] processing image 105/105 - /srv/dataset/jerry_seinfeld/httpsuploadwikimediaorgwikipediacommonsthumbbJerrySeinfeldjpgpxJerrySeinfeldjpg.jpg
[INFO] serializing encodings...
This app can be built and run as any other alwaysAI application. For general information on running apps, please see https://alwaysai.co/blog/building-and-deploying-apps-on-alwaysai
% aai app install
⚠ Application has no models
✔ Install python virtual environment
✔ Install python dependencies
% aai app start
status loading encodings + face detector...
Images:
['images/coarse_tilt_aligned_face.628.12094218513_3d47daa67b_o.jpg', 'images/coarse_tilt_aligned_face.722.12067405894_be037b3c25_o.jpg', 'images/coarse_tilt_aligned_face.723.12066554454_3bc03e02fd_o.jpg', 'images/httpabsolumentgratuitfreefrimagesbenaffleckjpg.jpg', 'images/httpafilesbiographycomimageuploadcfillcssrgbdprgfacehqwMTENDgMDUODczNDcNTcjpg.jpg', 'images/httpafilesbiographycomimageuploadcfillcssrgbdprgfacehqwMTIOTcwODQNTUzNjQMzcjpg.jpg', 'images/httpafilesbiographycomimageuploadcfillcssrgbdprgfacehqwMTIwNjANjMMDINzIxNjcjpg.jpg', 'images/httpaurorasblogcomwpcontentuploadsjerryseinfeldpublicityshotjpg.jpg', 'images/httpcdncdnjustjaredcomwpcontentuploadsheadlineseltonjohnsupportsbrucejennerstransitiontowomanjpg.jpg', 'images/httpcdncdnjustjaredcomwpcontentuploadsheadlinesmadonnatalksparisattackstearsjpg.jpg', 'images/httpcdncdnjustjaredcomwpcontentuploadsheadlinesmindykalingcomedypilotjpg.jpg', 'images/httpcdnfuncheapcomwpcontentuploadsVOGUEjpg.jpg', 'images/httpcdnlyricssongonlyricsnetwpcontentuploadsEltonJohnDiscographyCDreleasesjpg.jpg', 'images/httpcdnpastemagazinecomwwwarticlesmindykalingndbookjpg.jpg', 'images/httpecximagesamazoncomimagesIfmaBKWLACULSRjpg.jpg']
[INFO] Streamer started at http://localhost:5000
Open the link in a browser on your machine to start the facial recognition app.