add face bounding box

README.md

@@ -12,6 +12,8 @@ Please read the results as follow:
 
 Please download the DLIB [Shape Predictor](https://huggingface.co/matt3ounstable/dlib_predictor_recognition/resolve/main/shape_predictor_68_face_landmarks.dat?download=true) and the [Face Recognition](https://huggingface.co/matt3ounstable/dlib_predictor_recognition/resolve/main/dlib_face_recognition_resnet_model_v1.dat?download=true) models and place them into the `dlib` directory.
 
+Precompiled Dlib for windows can be found [here](https://github.com/z-mahmud22/Dlib_Windows_Python3.x).
+
 In this repository you also find a workflow that uses IPAdapter to generate a few images and return the distance to the reference face.
 
 ![face analysis](./face_analysis.jpg)

faceanalysis.py

@@ -59,9 +59,97 @@ class FaceAnalysisModels:
                 "shape_predict": dlib.shape_predictor(os.path.join(DLIB_DIR, "shape_predictor_68_face_landmarks.dat")),
                 "face_recog": dlib.face_recognition_model_v1(os.path.join(DLIB_DIR, "dlib_face_recognition_resnet_model_v1.dat")),
             }
+        
 
         return (out, )
 
+def crop_face(image, x, y, w, h, padding=0):
+    x = max(0, x - padding)
+    y = max(0, y - padding)
+    w = min(image.width, w + 2 * padding)
+    h = min(image.height, h + 2 * padding)
+
+    return image.crop((x, y, x + w, y + h))
+
+class FaceBoundingBox:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "analysis_models": ("ANALYSIS_MODELS", ),
+                "image": ("IMAGE", ),
+                "padding": ("INT", { "default": 0, "min": 0, "max": 4096, "step": 1 }),
+                "index": ("INT", { "default": -1, "min": -1, "max": 4096, "step": 1 }),
+            },
+        }
+
+    RETURN_TYPES = ("IMAGE", "INT", "INT", "INT", "INT")
+    RETURN_NAMES = ("IMAGE", "x", "y", "width", "height")
+    FUNCTION = "bbox"
+    CATEGORY = "FaceAnalysis"
+
+    def bbox(self, analysis_models, image, padding, index=-1):
+        out_img = []
+        out_x = []
+        out_y = []
+        out_w = []
+        out_h = []
+
+        for i in image:
+            img = T.ToPILImage()(i.permute(2, 0, 1)).convert('RGB')
+
+            if analysis_models["library"] == "insightface":
+                faces = analysis_models["detector"].get(np.array(img))
+                for face in faces:
+                    x, y, w, h = face.bbox.astype(int)
+                    w = w - x
+                    h = h - y
+                    x = max(0, x - padding)
+                    y = max(0, y - padding)
+                    w = min(img.width, w + 2 * padding)
+                    h = min(img.height, h + 2 * padding)
+                    crop = img.crop((x, y, x + w, y + h))
+                    out_img.append(T.ToTensor()(crop).permute(1, 2, 0))
+                    out_x.append(x)
+                    out_y.append(y)
+                    out_w.append(w)
+                    out_h.append(h)
+
+            else:
+                faces = analysis_models["detector"](np.array(img), 1)
+                for face in faces:
+                    x, y, w, h = face.left(), face.top(), face.width(), face.height()
+                    x = max(0, x - padding)
+                    y = max(0, y - padding)
+                    w = min(img.width, w + 2 * padding)
+                    h = min(img.height, h + 2 * padding)
+                    crop = img.crop((x, y, x + w, y + h))
+                    out_img.append(T.ToTensor()(crop).permute(1, 2, 0))
+                    out_x.append(x)
+                    out_y.append(y)
+                    out_w.append(w)
+                    out_h.append(h)
+
+        if not out_img:
+            raise Exception('No face detected in image.')
+                
+        out_img = torch.stack(out_img)
+
+        if out_img.shape[0] == 1:
+            index = 0
+
+        if index > out_img.shape[0] - 1:
+            index = out_img.shape[0] - 1
+
+        if index != -1:
+            out_img = out_img[index].unsqueeze(0)
+            out_x = out_x[index]
+            out_y = out_y[index]
+            out_w = out_w[index]
+            out_h = out_h[index]
+        
+        return (out_img, out_x, out_y, out_w, out_h,)
+
 class FaceEmbedDistance:
     @classmethod
     def INPUT_TYPES(s):
@@ -173,9 +261,11 @@ class FaceEmbedDistance:
 NODE_CLASS_MAPPINGS = {
     "FaceEmbedDistance": FaceEmbedDistance,
     "FaceAnalysisModels": FaceAnalysisModels,
+    "FaceBoundingBox": FaceBoundingBox,
 }
 
 NODE_DISPLAY_NAME_MAPPINGS = {
     "FaceEmbedDistance": "Face Embeds Distance",
     "FaceAnalysisModels": "Face Analysis Models",
+    "FaceBoundingBox": "Face Bounding Box",
 }