first commit

guillain · guillain · commit 28940c2cbe21 · 2019-12-29T17:49:53.000Z
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,5 @@
+.idea
+*.log
+*.json
+*pyc*
+.env
diff --git a/Dockerfile b/Dockerfile
@@ -0,0 +1,13 @@
+FROM jjanzic/docker-python3-opencv:latest
+# RPI: FROM sgtwilko/rpi-raspbian-opencv:latest
+
+WORKDIR /src
+
+COPY requirements.txt /src/requirements.txt
+
+RUN pip install --upgrade pip && \
+    pip install -r requirements.txt
+
+COPY src/ /src
+
+CMD ["/bin/bash"]
diff --git a/Dockerfile-rpi b/Dockerfile-rpi
@@ -0,0 +1,13 @@
+FROM sgtwilko/rpi-raspbian-opencv:latest
+#FROM mohaseeb/raspberrypi3-python-opencv:latest
+
+WORKDIR /src
+
+COPY requirements.txt /src/requirements.txt
+
+RUN pip install --upgrade pip && \
+    pip install -r requirements.txt
+
+COPY src/ /src
+
+CMD ["/bin/bash"]
diff --git a/LICENSE b/LICENSE
@@ -0,0 +1,5 @@
+Copyright (c) 2019 Guillain Sanchez <guillain@gmail.com>
+
+Private License
+
+Distribution not allowed
diff --git a/README.md b/README.md
@@ -0,0 +1,32 @@
+# gPhoto-object-extraction
+_How to run easily the object recognition_
+
+For the demo and to make it more funny:
+- the source of information (means the photo galery) is *Google Photo*!
+- the object recognized are stored on *AWS S3*!
+- all run in docker :)
+- ... compatible with Raspberry 
+
+## Pre-requisite
+- Docker host
+- Internet access
+
+## Easy Run
+After have cloned the repository localy (`git clone https://github.com/guillain/gPhoto-object-extraction`)
+and enter in the new directory (`cd gPhoto-object-extraction`) execute the following commands:
+
+1/ create your own confguration file: 
+  - `cp sample.env .env && vi .env` and add your AWS credentials info
+  - import the files `google-photo.json` and `google-photo-token.json` as GCP credentials info
+  - if you need to run it on Raspberry, invert the comment on the first two lines in the `Dockerfile`
+
+2/ execute the container with the desired date for your Google Photos: `./run "29/12/2019"`
+
+
+## Support
+- Original post:
+  - https://www.hackster.io/mjrobot/real-time-face-recognition-an-end-to-end-project-a10826
+  - https://github.com/Mjrovai/OpenCV-Face-Recognition
+- Raspberry: https://www.raspberrypi.org
+- Docker: https://www.docker.com
+- OpenCV: https://opencv.org
diff --git a/requirements.txt b/requirements.txt
@@ -0,0 +1,10 @@
+tensorflow
+python-forecastio
+opencv-python==4.1.0.25
+awscli
+boto3
+numpy
+google-api-python-client
+httplib2
+oauth2client
+python-dotenv
diff --git a/run b/run
@@ -0,0 +1,32 @@
+#!/bin/bash
+. .env
+
+# my_date=${1} if not provided, go in bash
+
+# Name of the container (to build & run)
+NAME="gphoto-object-extraction"
+
+# If no date, go in console
+if [ "${1}" == "" ]; then
+  echo -e '\n/!\ No date provided, we jump in the console (exit to go out) /!\ \n'
+  RUN="/bin/bash"
+else
+  RUN="python app.py"
+fi
+
+# Build the image
+docker build -t ${NAME} .
+
+# Test the image and the cv2 import
+docker run -it --rm ${NAME} python -c "import cv2; print(cv2.__version__)"
+
+# Run the app
+docker run -it --rm \
+  -e my_date=${1} \
+  --env-file=`pwd`/.env \
+  -v `pwd`/google-photo.json:/src/google-photo.json \
+  -v `pwd`/google-photo-token.json:/src/google-photo-token.json \
+  ${NAME} ${RUN}
+
+# Done & Bye
+exit 0
diff --git a/sample.env b/sample.env
@@ -0,0 +1,7 @@
+aws_s3_bucket=my_obects
+aws_region_name=eu-central-1
+aws_key_id=
+aws_access_key=
+
+gcp_ident_file=google-photo.json
+gcp_token_file=google-photo-token.json
diff --git a/setup.py b/setup.py
@@ -0,0 +1,22 @@
+from setuptools import setup, find_packages
+
+NAME="gPhoto-object-extraction"
+__version__ = '1.0.0'
+AUTHOR="Guillain"
+AUTHOR_EMAIL="guillain@gmail.com"
+
+with open('requirements.txt') as f:
+    requirements = f.read().splitlines()
+
+setup(
+    name=NAME,
+    version=__version__,
+    description="Image analysis with IA",
+    url='https://gitlab.com/bo-art-of-bonsai/' + NAME,
+    author=AUTHOR,
+    author_email=AUTHOR_EMAIL,
+    license='Private',
+    install_requires=requirements,
+    packages=find_packages(),
+    zip_safe=False
+)
diff --git a/src/app.py b/src/app.py
@@ -0,0 +1,69 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8
+import os
+import requests
+
+from lib.aws import AWS
+from lib.opencv import OpenCV
+from lib.gcp import GCP
+
+from dotenv import load_dotenv
+load_dotenv()
+
+
+def main(my_date):
+    # my_date = "29/11/2019"
+
+    # Instancies
+    aws = AWS(bucket=os.getenv('aws_s3_bucket'))
+    gcp = GCP(cred_file=os.getenv('gcp_token_file'))
+    cv = OpenCV()
+
+    # Get list of photos
+    photos = gcp.listMedia(date_filter=my_date, media_type=["PHOTO"])
+    if photos is None:
+        print('error', 'No photos found, exit')
+        exit(0)
+    print('{} photos found'.format(len(photos)))
+
+    # For each photo: download, extract each face and store individual face in S3
+    face_catalog = []
+    for photo in photos:
+        # Download photo data
+        photo_data = requests.get(photo['baseUrl']).content
+        with open(photo['filename'], 'wb') as handler:
+            handler.write(photo_data)
+
+        faces = cv.extract_faces(photo['filename'])
+        #faces = cv.extract_faces_from_buffer(photo_data)
+
+        i_face = 1
+        for face in faces:
+            filename = 'extract/{}'.format(face)
+
+            aws.s3_upload_file(face, filename)
+            os.remove(face)
+            #aws.s3_upload_data(face, filename)
+
+            face = dict(photo)
+            face['face_id'] = i_face
+            face['face_filename'] = filename
+            #del face['baseUrl']
+            face_catalog.append(face)
+
+            i_face = i_face + 1
+
+        os.remove(photo['filename'])
+        print('{} face(s) found in the photo {}'.format(i_face - 1, photo['filename']))
+
+    if face_catalog:
+        for face in face_catalog:
+            print('JSON catalog of faces\n{}'.format(face))
+        print('{} face(s) found in total'.format(len(face_catalog)))
+    else:
+        print('No face found :(')
+
+
+if __name__ == '__main__':
+    main(os.getenv('my_date'))
+
diff --git a/src/lib/aws.py b/src/lib/aws.py
@@ -0,0 +1,42 @@
+import os
+import boto3
+from io import BytesIO
+
+
+class AWS(object):
+    def __init__(self, 
+               type='s3',
+               region=os.getenv('aws_region_name'),
+               key_id=os.getenv('aws_key_id'),
+               access_key=os.getenv('aws_access_key'),
+               bucket='bo-infra-backup'):
+        self.region_name = region
+        self.bucket_name = bucket
+
+        self.session = boto3.Session(
+            aws_access_key_id=key_id,
+            aws_secret_access_key=access_key)
+
+    def init(self, type='s3', obj='resource'):
+        if obj == 'resource':
+            return self.session.resource(type, region_name=self.region_name)
+        elif obj == 'client':
+            return self.session.client(type, region_name=self.region_name)
+
+    def s3_download(self, file, output):
+        s3 = self.init(type='s3', obj='resource')
+        file_stream = BytesIO()
+
+        s3.Object(self.bucket_name, file).download_file(output)
+        s3.Object(self.bucket_name, file).download_fileobj(file_stream)
+
+    def s3_upload_file(self, file_in, file_out):
+        s3 = self.init(type='s3', obj='resource')
+
+        s3.Bucket(self.bucket_name).upload_file(Filename=file_in, Key=file_out)
+
+    def s3_upload_data(self, data, filename):
+        s3 = self.init(type='s3', obj='resource')
+
+        s3.Object(self.bucket_name, filename).put(Body=data)
+
diff --git a/src/lib/gcp.py b/src/lib/gcp.py
@@ -0,0 +1,74 @@
+import os
+from googleapiclient.discovery import build, MediaFileUpload
+from httplib2 import Http
+from oauth2client import file, client, tools
+
+
+class GCP(object):
+    def __init__(self, cred_file, store_file=os.getenv('gcp_ident_file')):
+        self.store_file = store_file
+        # self.photo_scopes = 'https://www.googleapis.com/auth/photoslibrary.readonly'
+        self.photo_scopes = 'https://www.googleapis.com/auth/photoslibrary.sharing'
+
+        store = file.Storage(self.store_file)
+        creds = store.get()
+        if not creds or creds.invalid:
+            flow = client.flow_from_clientsecrets(cred_file, self.photo_scopes)
+            creds = tools.run_flow(flow, store)
+
+        self.lib = build('photoslibrary', 'v1', http=creds.authorize(Http()))
+
+    def format_date(self, my_date):
+        # my_date = "27/12/2019"
+
+        my_date_arr = my_date.split('/')
+        return [{"day": my_date_arr[0], "month": my_date_arr[1], "year": my_date_arr[2]}]
+
+    def listMedia(self, date_filter=None, media_type=["PHOTO", "VIDEO"]):
+        # date_filter = "27/12/2019" or "0/12/2019" for all days of december
+
+        nextpagetoken = 'Dummy'
+        filters = { "mediaTypeFilter": {"mediaTypes": media_type} }
+        if date_filter is not None:
+            filters = {
+                    "dateFilter": {"dates": self.format_date(date_filter)},
+                    "mediaTypeFilter": {"mediaTypes": media_type}
+            }
+
+        res = []
+        while nextpagetoken != '':
+            nextpagetoken = '' if nextpagetoken == 'Dummy' else nextpagetoken
+            results = self.lib.mediaItems().search(
+                body={
+                    "filters":  filters, 
+                    "pageSize": 100, 
+                    "pageToken": nextpagetoken,
+                }
+            ).execute()
+                          
+            # The default number of media items to return at a time is 25. The maximum pageSize is 100.
+            items = results.get('mediaItems', [])
+            nextpagetoken = results.get('nextPageToken', '')
+            for item in items:
+                res.append(item)
+                
+        return res
+    
+    def uploadMedia(self, media, name='', mimetype='image/png'):
+        media = MediaFileUpload(media, mimetype=mimetype, resumable=True)
+        request = self.lib.create(media_body=media, body={'name': name})
+        response = None
+        while response is None:
+            status, response = request.next_chunk()
+
+        return response
+
+    def listAlbums(self):
+        r = self.lib.albums().list(pageSize=10).execute()
+        items = r.get('albums', [])
+        return items
+    
+    def createAlbum(self, title):
+        r = self.lib.albums().create(body={'album':{'title':title}}).execute()
+        return r.id
+
diff --git a/src/lib/haarcascade_frontalface_default.xml b/src/lib/haarcascade_frontalface_default.xml
diff --git a/src/lib/opencv.py b/src/lib/opencv.py

-Original file line number
+Diff line change
@@ @@ -0,0 +1,5 @@ @@
 +.idea
 +*.log
 +*.json
 +*pyc*
 +.env