Added webcam demo for Face Recognition (#247)

* Added webcam demo for Face Recognition

* Update webcam_demo.py

* Renamed README.md and added webcam_demo instructions.

* Renamed README.md and added webcam_demo instructions.

* Renamed README.md and added webcam_demo instructions.

* Renamed README.md and added webcam_demo instructions.

* Renamed README.md and added webcam_demo instructions.

* Renamed README.md and added webcam_demo instructions.

Co-authored-by: ad-daniel <[email protected]>

Author: Pavlos-Tosidis

Diff for: projects/perception/face_recognition/REAMDE.md renamed to projects/perception/face_recognition/README.md

@@ -8,5 +8,7 @@ More specifically, the following applications are provided:
 2. demos/eval_demo.py: A tool that demonstrates how to perform evaluation using FaceRecognition
 3. demos/inference_demo.py: A tool that demonstrates how to perform inference on a single image
 4. demos/benchmarking_demo.py: A simple benchmarking tool for measuring the performance of FaceRecognition in various platforms
+5. demos/webcam_demo.py: A tool that demonstrates how to perform face detection and recognition with the use of a webcam.
+   1. To use this tool you have to first create a database containing the faces to be recognised. To do this, you will have to prepare the face images using the [align](https://github.com/opendr-eu/opendr/blob/master/docs/reference/face-recognition.md#facerecognitionlearneralign) method of the tool and place them in a folder named `'cropped_images_path'` inside the `'demos'` directory.
 
 Please use the --device cpu flag for the demos if you are running them on a machine without a CUDA-enabled GPU.

Diff for: projects/perception/face_recognition/demos/webcam_demo.py

@@ -0,0 +1,63 @@
+# Copyright 2020-2022 OpenDR European Project
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+import cv2
+
+from opendr.perception.object_detection_2d import RetinaFaceLearner
+from opendr.perception.object_detection_2d.datasets.transforms import\
+    BoundingBoxListToNumpyArray
+from opendr.perception.face_recognition import FaceRecognitionLearner
+
+facedetector = RetinaFaceLearner(backbone='mnet', device='cuda')
+facedetector.download(".", mode="pretrained")
+facedetector.load("./retinaface_mnet")
+
+recognizer = FaceRecognitionLearner(device='cuda', backbone='mobilefacenet', mode='backbone_only')
+recognizer.download(path=".")
+recognizer.load(".")
+recognizer.fit_reference('./cropped_images_path', save_path="./save_path", create_new=True)
+
+cam = cv2.VideoCapture(0)
+cv2.namedWindow("face recognition")
+font = cv2.FONT_HERSHEY_SIMPLEX
+fontScale = 1
+color = (255, 0, 0)
+thickness = 2
+while True:
+    ret, frame = cam.read()
+    if not ret:
+        print("Failed to grab frame")
+        break
+    bounding_boxes = facedetector.infer(frame)
+    if bounding_boxes:
+        bounding_boxes_ = BoundingBoxListToNumpyArray()(bounding_boxes)
+        boxes = bounding_boxes_[:, :4]
+        for idx, box in enumerate(boxes):
+            (startX, startY, endX, endY) = int(box[0]), int(box[1]), int(box[2]), int(box[3])
+            img = frame[startY:endY, startX:endX]
+            result = recognizer.infer(img)
+            if result.description != 'Not found':
+                color = (0, 255, 0)
+            else:
+                color = (0, 0, 255)
+            img = cv2.rectangle(frame, (startX, startY), (endX, endY), color, thickness)
+            img = cv2.putText(img, result.description, (startX, endY - 10), font,
+                              fontScale, color, thickness, cv2.LINE_AA)
+    else:
+        img = frame
+    cv2.imshow("face recognition", img)
+    cv2.waitKey(1)
+
+cam.release()