make cli a mvp

src/protestswap/cli.py

@@ -1,6 +1,10 @@
 import os
 import logging
 
+import argparse
+
+import cv2
+
 from protestswap.protestswap import ProtestFaceSwapper
 
 logger = logging.getLogger("protestswap")
@@ -10,19 +14,47 @@ INSIGHTFACE_VAR = "INSIGHTFACE_ROOT_DIR"
 
 
 def main():
+    parser = argparse.ArgumentParser(
+        prog="protestswap", description="Swaps faces of people on protests"
+    )
+    parser.add_argument("target", help="The image to swap the faces on")
+    parser.add_argument(
+        "-t", "--template", action="append", help="The template faces to paste on"
+    )
+
+    args = parser.parse_args()
+
+    print(args.target)
+
+    if not args.template:
+        print("You need at least one template (with '-t')!")
+        exit(1)
+
     insightface_dir = None
-    if os.path.exists(DEFAULT_INSIGHTFACE_DIR):
-        insightface_dir = DEFAULT_INSIGHTFACE_DIR
-    elif os.environ.get(INSIGHTFACE_VAR):
+    if os.environ.get(INSIGHTFACE_VAR):
         insightface_dir = os.environ.get(INSIGHTFACE_VAR)
+    elif os.path.exists(DEFAULT_INSIGHTFACE_DIR):
+        insightface_dir = DEFAULT_INSIGHTFACE_DIR
     else:
         logger.warning(
             f"No directory at '{DEFAULT_INSIGHTFACE_DIR}' and '{INSIGHTFACE_VAR}' not set yet."
         )
         insightface_dir = f"{os.environ.get('HOME')}/.cache/insightface"
 
-    # swapper = ProtestFaceSwapper(insightface_dir)
-    # swapper.prepare_model()
+    swapper = ProtestFaceSwapper(insightface_dir)
+    swapper.prepare_model()
+
+    swapper.set_target_path(args.target)
+
+    for template in args.template:
+        print(f"Adding template {template}...")
+        swapper.add_template_path(template)
+
+    print("Detecting faces")
+    swapper.detect_target_faces()
+
+    swapped = swapper.swap()
+    cv2.imwrite("result.png", swapped)
 
 
 if __name__ == "__main__":

src/protestswap/protestswap.py

@@ -1,10 +1,10 @@
 #!/usr/bin/env python3
 
-import logging
 import os
+import logging
 
 import cv2
-import random # TODO: seed!
+import random  # TODO: seed!
 
 import insightface
 from insightface.app import FaceAnalysis
@@ -14,51 +14,65 @@ from insightface.app.common import Face
 
 logger = logging.getLogger("protestswap")
 
-SWAPPING_MODEL = 'inswapper_128.onnx'
+SWAPPING_MODEL = "inswapper_128.onnx"
 
 
 class ProtestFaceSwapper:
     def __init__(self, root_dir):
-        self._app : FaceAnalysis = FaceAnalysis(name='buffalo_l', root=insightface_dir)
-        self._model : INSwapper
-        self._source : cv2.typing.MatLike|None = None
-        self._source_faces : list[Face] = []
-        self._template_faces : list[Face] = []
+        self._root_dir = root_dir
+        self._app: FaceAnalysis = FaceAnalysis(name="buffalo_l", root=root_dir)
+        self._model: INSwapper
+        self._target: cv2.typing.MatLike | None = None
+        self._target_faces: list[Face] = []
+        self._template_faces: list[Face] = []
 
         self._app.prepare(ctx_id=0, det_size=(640, 640))
 
     def prepare_model(self) -> None:
-        self._model = insightface.model_zoo.get_model(SWAPPING_MODEL, download=True, download_zip=True) # Revealed type: "INSwapper"
+        self._model = insightface.model_zoo.get_model(
+            os.path.join(self._root_dir, "models", SWAPPING_MODEL),
+            root=self._root_dir,
+            download=False,
+            download_zip=False,
+        )  # Revealed type: "INSwapper"
+        print(f"Model: {self._model}")
 
-    def set_source(self, path: str) -> None:
-        self._source = cv2.imread(path)
+    def set_target_path(self, path: str) -> None:
+        self.set_target(cv2.imread(path))
 
-    def detect_source_faces(self):
-        self._source_faces = self._app.get(self._source)
+    def set_target(self, image: cv2.UMat) -> None:
+        self._target = image
 
-    def add_template_faces(self, template_path: str):
-        self._template_faces += [self._app.get(cv2.imread(template_path))]
+    def detect_target_faces(self):
+        self._target_faces = self._app.get(self._target)
+        print(f"Found {len(self._target_faces)} faces!")
+
+    def add_template_path(self, template_path: str):
+        self.add_template(cv2.imread(template_path))
+
+    def add_template(self, template: cv2.UMat):
+        self._template_faces += self._app.get(template)
 
     def swap(self) -> cv2.typing.MatLike:
-        work_copy = self._source_faces.copy()
-        for i, face in enumerate(self._source_faces):
-            print(f"Replacing face {i+1}/{len(self._source_faces)}...", end="", flush=True);
+        work_copy = self._target.copy()
+        for i, face in enumerate(self._target_faces):
+            print(
+                f"Replacing face {i + 1}/{len(self._target_faces)}...",
+                # end="",
+                # flush=True,
+            )
             imposed_face = random.choice(self._template_faces)
-            work_copy = self._model.get(work_copy, face, imposed_face, paste_back=True)
+            print(
+                f"work_copy: {type(work_copy)}, face: {type(face)}, imposed: {
+                    type(imposed_face)
+                }"
+            )
+            work_copy = self._model.get(
+                work_copy, face, imposed_face, paste_back=True)
             print(" done!")
 
         return work_copy
 
     @staticmethod
-    def readFile(path: str) -> cv2.typing.MatLike|None:
+    def readFile(path: str) -> cv2.typing.MatLike | None:
         return cv2.imread(path)
-
-
-def main():
-
-    swapper = ProtestFaceSwapper(insightface_dir)
-    swapper.prepare_model()
-
-
-if __name__ == '__main__':
-    main()