Initial commit.

.gitignore

@@ -0,0 +1,15 @@
+
+# Ignore anything
+*
+
+# Except '.gitignore'
+!.gitignore
+
+# Except 'README.md'
+!README.md
+
+# Except '*.py'
+!*.py
+
+# End-of-file
+

README.md

@@ -0,0 +1,3 @@
+
+# ComfyUI-ImageCropper
+

main.py

@@ -0,0 +1,148 @@
+
+import cv2
+import PIL
+import numpy as np
+import os
+import torch
+
+mode2config = {
+    "eye (with eye-glasses)":        "haarcascade_eye_tree_eyeglasses.xml",
+    "eye":                           "haarcascade_eye.xml",
+    "left eye (2 splits)":           "haarcascade_lefteye_2splits.xml",
+    "right eye (2 splits)":          "haarcascade_righteye_2splits.xml",
+    "frontal face (extended)":       "haarcascade_frontalcatface_extended.xml",
+    "frontal cat face":              "haarcascade_frontalcatface.xml",
+    "frontal face (alternate 2)":    "haarcascade_frontalface_alt2.xml",
+    "frontal face (alternate tree)": "haarcascade_frontalface_alt_tree.xml",
+    "frontal face (alternate 1)":    "haarcascade_frontalface_alt.xml",
+    "frontal face (default)":        "haarcascade_frontalface_default.xml",
+    "full body":                     "haarcascade_fullbody.xml",
+    "lower body":                    "haarcascade_lowerbody.xml",
+    "upper body":                    "haarcascade_upperbody.xml",
+    "profile face":                  "haarcascade_profileface.xml",
+    "smile":                         "haarcascade_smile.xml",
+    "license plate number":          "haarcascade_license_plate_rus_16stages.xml",
+    "russian plate number":          "haarcascade_russian_plate_number.xml",
+}
+
+def tensor2pil(image):
+    return PIL.Image.fromarray(np.clip(255. * image.cpu().numpy().squeeze(0), 0, 255).astype(np.uint8))
+
+def pil2tensor(image):
+    return torch.from_numpy(np.array(image).astype(np.float32) / 255.0).unsqueeze(0)
+
+def clamp(xmin, xval, xmax):
+    if xval < xmin:
+        return xmin
+    if xval > xmax:
+        return xmax
+    return xval
+
+def cropImage(image:PIL.Image, x1, y1, x2, y2):
+    mw, mh = image.size
+    return image.crop((
+        clamp(0, x1, mw),
+        clamp(0, y1, mh),
+        clamp(0, x2, mw),
+        clamp(0, y2, mh)))
+
+class ImageCropper:
+   
+    def __init__(self):
+        pass
+ 
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "image": ("IMAGE",),
+                "mode": ([
+                    #"eye (with eye-glasses)",
+                    #"eye",
+                    #"left eye (2 splits)",
+                    #"right eye (2 splits)",
+                    "frontal face (extended)",
+                    #"frontal cat face",
+                    #"frontal face (alternate 2)",
+                    #"frontal face (alternate tree)",
+                    #"frontal face (alternate 1)",
+                    "frontal face (default)",
+                    #"full body",
+                    #"lower body",
+                    #"upper body",
+                    "profile face",
+                    #"smile",
+                    #"license plate number",
+                    #"russian plate number",
+                ],),
+                "padding": ("INT", {
+                    "default": 0, 
+                    "min": 0, #Minimum value
+                    "step": 1, #Slider's step
+                    "display": "number" # Cosmetic only: display as "number" or "slider"
+                }),
+            },
+        }
+
+    # Configuration for the classifier can be found at:
+    #   https://github.com/opencv/opencv/tree/master/data/haarcascades
+    def parseMode(self, mode):
+        folder = os.path.dirname(__file__)
+        folder = os.path.join(folder, "classifiers")
+        return os.path.join(folder, mode2config[mode])
+
+    RETURN_TYPES = ("IMAGE",)
+    FUNCTION = "execute"
+    CATEGORY = "utils"
+
+    def execute(self, image, mode, padding):
+
+        print(f"Input mode: {mode}")
+
+        config = self.parseMode(mode)
+
+        classifier = cv2.CascadeClassifier(config)
+
+        pilImage = tensor2pil(image)
+
+        converted = cv2.cvtColor(
+            np.array(pilImage),
+            cv2.COLOR_RGB2GRAY)
+ 
+        founds = classifier.detectMultiScale(
+            converted, 
+            scaleFactor=1.1,
+            minNeighbors=5, 
+            minSize=(30, 30), 
+            flags=cv2.CASCADE_SCALE_IMAGE)
+
+        count = len(founds) 
+        if count == 0:
+            raise Exception(f"Found no matched")
+        if count > 1:
+            raise Exception(f"Found multiple matched")
+
+        (x, y, w, h) = founds[0]
+        padded = cropImage(
+            pilImage,
+            x-padding,
+            y-padding,
+            x+w+padding,
+            y+h+padding) 
+        return (pil2tensor(padded),)
+
+    #@classmethod
+    #def IS_CHANGED(s, image, mode):
+    #    return ""
+
+# A dictionary that contains all nodes you want to export with their names
+# NOTE: names should be globally unique
+NODE_CLASS_MAPPINGS = {
+    "ImageCropper": ImageCropper
+}
+
+# A dictionary that contains the friendly/humanly readable titles for the nodes
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "ImageCropper": "Image cropping tool"
+}
+