Added facial recognition confidence threshold

Also reset version and added release workflow

.github/workflows/python-publish.yml

@@ -0,0 +1,39 @@
+# This workflow will upload a Python Package using Twine when a release is created
+# For more information see: https://docs.github.com/en/actions/automating-builds-and-tests/building-and-testing-python#publishing-to-package-registries
+
+# This workflow uses actions that are not certified by GitHub.
+# They are provided by a third-party and are governed by
+# separate terms of service, privacy policy, and support
+# documentation.
+
+name: Upload Python Package
+
+on:
+  release:
+    types: [published]
+
+permissions:
+  contents: read
+
+jobs:
+  deploy:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v3
+    - name: Set up Python
+      uses: actions/setup-python@v3
+      with:
+        python-version: '3.x'
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install build
+    - name: Build package
+      run: python -m build
+    - name: Publish package
+      uses: pypa/gh-action-pypi-publish@v1.8.10
+      with:
+        user: __token__
+        password: ${{ secrets.PYPI_PROJECT_API_TOKEN }}

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "wyzely-detect"
-version = "0.1.9"
+version = "0.1.0"
 description = "Recognize faces/objects in a video stream (from a webcam or a security camera) and send notifications to your devices"
 authors = ["slashtechno <77907286+slashtechno@users.noreply.github.com>"]
 license = "MIT"
@@ -26,10 +26,10 @@ torch = ">=2.0.0, !=2.0.1, !=2.1.0"
 # https://discuss.tensorflow.org/t/tensorflow-io-gcs-filesystem-with-windows/18849/4
 # Might be able to remove this version constraint later
 tensorflow-io-gcs-filesystem = "0.31.0"
+tensorflow = "^2.14.0"
 
 
 deepface = "^0.0.79"
-tensorflow = "^2.14.0"
 
 [tool.poetry.group.dev.dependencies]
 black = "^23.9.1"

wyzely_detect/utils/utils.py

@@ -74,6 +74,7 @@ def recognize_face(
     In addition, accepts an opencv image to be used as the frame to be searched
 
     Returns a single dictonary as currently only 1 face can be detected in each frame
+    Cosine threshold is 0.3, so if the confidence is less than that, it will return None
     dict contains the following keys: label, x1, y1, x2, y2
     The directory should be structured as follows:
     faces/
@@ -101,11 +102,6 @@ def recognize_face(
             print("representations_arcface.pkl does not exist")
         first_face_try = False
 
-    # For debugging
-    # if path_to_directory.joinpath("representations_arcface.pkl").exists():
-    #     print("representations_arcface.pkl exists")
-    # else:
-    #     print("representations_arcface.pkl does not exist")
 
     # face_dataframes is a vanilla list of dataframes
     # It seems face_dataframes is empty if the face database (directory) doesn't exist. Seems to work if it's empty though
@@ -115,8 +111,8 @@ def recognize_face(
         face_dataframes = DeepFace.find(
             run_frame,
             db_path=str(path_to_directory),
-            # enforce_detection=True,
-            # Seems this works?
+            # Problem with enforce_detection=False is that it will always (?) return a face, no matter the confidence
+            # Thus, false-positives need to be filtered out
             enforce_detection=False,
             silent=True,
             # Could use VGG-Face, but whilst fixing another issue, ArcFace seemed to be slightly faster
@@ -152,9 +148,11 @@ def recognize_face(
             "x2": df.iloc[-1]["source_x"] + df.iloc[-1]["source_w"],
             "y2": df.iloc[-1]["source_y"] + df.iloc[-1]["source_h"],
         }
-        # After some brief testing, it seems positve matches are > 0.3
-        # I have not seen any false positives, so there is no threashold yet
+        # After some brief testing, it seems positive matches are > 0.3
         distance = df.iloc[-1]["ArcFace_cosine"]
+        # TODO: Make this a CLI argument
+        if distance < 0.3:
+            return None
         # if 0.5 < distance < 0.7:
         # label = "Unknown"
         to_return = dict(label=label, **coordinates)