Doron Adler commited on
Commit
d98412d
1 Parent(s): c63f589

Head Pumpkinizer

Browse files
.gitattributes CHANGED
@@ -31,3 +31,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
31
  *.zip filter=lfs diff=lfs merge=lfs -text
32
  *.zst filter=lfs diff=lfs merge=lfs -text
33
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
31
  *.zip filter=lfs diff=lfs merge=lfs -text
32
  *.zst filter=lfs diff=lfs merge=lfs -text
33
  *tfevents* filter=lfs diff=lfs merge=lfs -text
34
+ ffhq2pumpkinheads_pix2pixHD_e03_itr4500-simp.onnx filter=lfs diff=lfs merge=lfs -text
35
+ shape_predictor_5_face_landmarks.dat filter=lfs diff=lfs merge=lfs -text
Example00001.jpg ADDED
Example00002.jpg ADDED
Example00003.jpg ADDED
Example00004.jpg ADDED
README.md CHANGED
@@ -1,13 +1,39 @@
1
  ---
2
- title: PumpkinHeads
3
- emoji: 🐨
4
- colorFrom: purple
5
- colorTo: indigo
6
  sdk: gradio
7
- sdk_version: 3.6
8
  app_file: app.py
9
  pinned: false
10
- license: mit
11
  ---
12
 
13
- Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ title: Pumpkin heads
3
+ emoji: 🎃
4
+ colorFrom: orange
5
+ colorTo: yellow
6
  sdk: gradio
 
7
  app_file: app.py
8
  pinned: false
 
9
  ---
10
 
11
+ Check out the configuration reference at https://huggingface.co/docs/hub/spaces#reference
12
+
13
+ # Configuration
14
+
15
+ `title`: _string_
16
+ Display title for the Space
17
+
18
+ `emoji`: _string_
19
+ Space emoji (emoji-only character allowed)
20
+
21
+ `colorFrom`: _string_
22
+ Color for Thumbnail gradient (red, yellow, green, blue, indigo, purple, pink, gray)
23
+
24
+ `colorTo`: _string_
25
+ Color for Thumbnail gradient (red, yellow, green, blue, indigo, purple, pink, gray)
26
+
27
+ `sdk`: _string_
28
+ Can be either `gradio` or `streamlit`
29
+
30
+ `sdk_version` : _string_
31
+ Only applicable for `streamlit` SDK.
32
+ See [doc](https://hf.co/docs/hub/spaces) for more info on supported versions.
33
+
34
+ `app_file`: _string_
35
+ Path to your main application file (which contains either `gradio` or `streamlit` Python code).
36
+ Path is relative to the root of the repository.
37
+
38
+ `pinned`: _boolean_
39
+ Whether the Space stays on top of your list.
app.py ADDED
@@ -0,0 +1,102 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import onnxruntime as rt
3
+ import sys
4
+ import PIL
5
+ from PIL import Image, ImageOps, ImageFile
6
+ import numpy as np
7
+ from pathlib import Path
8
+ import collections
9
+ from typing import Union, List
10
+ import scipy.ndimage
11
+ import requests
12
+
13
+ MODEL_FILE = "ffhq2pumpkinheads_pix2pixHD_e03_itr4500-simp.onnx"
14
+ so = rt.SessionOptions()
15
+ so.inter_op_num_threads = 4
16
+ so.intra_op_num_threads = 4
17
+ session = rt.InferenceSession(MODEL_FILE, sess_options=so)
18
+ input_name = session.get_inputs()[0].name
19
+ print("input_name = " + str(input_name))
20
+ output_name = session.get_outputs()[0].name
21
+ print("output_name = " + str(output_name))
22
+
23
+ import os
24
+ os.system("pip install dlib")
25
+ import face_detection
26
+
27
+ def array_to_image(array_in):
28
+ array_in = np.squeeze(255*(array_in + 1)/2)
29
+ array_in = np.transpose(array_in, (1, 2, 0))
30
+ im = Image.fromarray(array_in.astype(np.uint8))
31
+ return im
32
+
33
+ def image_as_array(image_in):
34
+ im_array = np.array(image_in, np.float32)
35
+ im_array = (im_array/255)*2 - 1
36
+ im_array = np.transpose(im_array, (2, 0, 1))
37
+ im_array = np.expand_dims(im_array, 0)
38
+ return im_array
39
+
40
+ def find_aligned_face(image_in, size=1024):
41
+ aligned_image, n_faces, quad = face_detection.align(image_in, face_index=0, output_size=size)
42
+ return aligned_image, n_faces, quad
43
+
44
+ def align_first_face(image_in, size=1024):
45
+ aligned_image, n_faces, quad = find_aligned_face(image_in,size=size)
46
+ if n_faces == 0:
47
+ try:
48
+ image_in = ImageOps.exif_transpose(image_in)
49
+ except:
50
+ print("exif problem, not rotating")
51
+ image_in = image_in.resize((size, size))
52
+ im_array = image_as_array(image_in)
53
+ else:
54
+ im_array = image_as_array(aligned_image)
55
+
56
+ return im_array
57
+
58
+ def img_concat_h(im1, im2):
59
+ dst = Image.new('RGB', (im1.width + im2.width, im1.height))
60
+ dst.paste(im1, (0, 0))
61
+ dst.paste(im2, (im1.width, 0))
62
+ return dst
63
+
64
+ import gradio as gr
65
+
66
+ def face2vintage(
67
+ img: Image.Image,
68
+ size: int
69
+ ) -> Image.Image:
70
+
71
+ aligned_img = align_first_face(img)
72
+ if aligned_img is None:
73
+ output=None
74
+ else:
75
+ output = session.run([output_name], {input_name: aligned_img})[0]
76
+ output = array_to_image(output)
77
+ aligned_img = array_to_image(aligned_img).resize((output.width, output.height))
78
+ output = img_concat_h(aligned_img, output)
79
+
80
+ return output
81
+
82
+ def inference(img):
83
+ out = face2vintage(img, 1024)
84
+ return out
85
+
86
+
87
+ title = "Pumpkin head Pix2PixHD"
88
+ description = "Pumpkinize your head. Upload an image with a face, or click on one of the examples below. If a face could not be detected, an image will still be created."
89
+ article = "<hr><p style='text-align: center'>See the <a href='https://github.com/justinpinkney/pix2pixHD' target='_blank'>Github Repo</a></p><p>The \"Pumpkin Head\" Pix2PixHD model was trained by <a href='https://linktr.ee/Norod78' target='_blank'>Doron Adler</a></p>"
90
+
91
+ examples=[['Example00001.jpg'],['Example00002.jpg'],['Example00003.jpg'],['Example00004.jpg']]
92
+ gr.Interface(
93
+ inference,
94
+ gr.inputs.Image(type="pil", label="Input"),
95
+ gr.outputs.Image(type="pil", label="Output"),
96
+ title=title,
97
+ description=description,
98
+ article=article,
99
+ examples=examples,
100
+ enable_queue=True,
101
+ allow_flagging=False
102
+ ).launch()
face_detection.py ADDED
@@ -0,0 +1,140 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021 Justin Pinkney
2
+
3
+ import dlib
4
+ import numpy as np
5
+ import os
6
+ from PIL import Image
7
+ from PIL import ImageOps
8
+ from scipy.ndimage import gaussian_filter
9
+ import cv2
10
+
11
+
12
+ MODEL_PATH = "shape_predictor_5_face_landmarks.dat"
13
+ detector = dlib.get_frontal_face_detector()
14
+
15
+
16
+ def align(image_in, face_index=0, output_size=256):
17
+ try:
18
+ image_in = ImageOps.exif_transpose(image_in)
19
+ except:
20
+ print("exif problem, not rotating")
21
+
22
+ landmarks = list(get_landmarks(image_in))
23
+ n_faces = len(landmarks)
24
+ face_index = min(n_faces-1, face_index)
25
+ if n_faces == 0:
26
+ aligned_image = image_in
27
+ quad = None
28
+ else:
29
+ aligned_image, quad = image_align(image_in, landmarks[face_index], output_size=output_size)
30
+
31
+ return aligned_image, n_faces, quad
32
+
33
+
34
+ def composite_images(quad, img, output):
35
+ """Composite an image into and output canvas according to transformed co-ords"""
36
+ output = output.convert("RGBA")
37
+ img = img.convert("RGBA")
38
+ input_size = img.size
39
+ src = np.array(((0, 0), (0, input_size[1]), input_size, (input_size[0], 0)), dtype=np.float32)
40
+ dst = np.float32(quad)
41
+ mtx = cv2.getPerspectiveTransform(dst, src)
42
+ img = img.transform(output.size, Image.PERSPECTIVE, mtx.flatten(), Image.BILINEAR)
43
+ output.alpha_composite(img)
44
+
45
+ return output.convert("RGB")
46
+
47
+
48
+ def get_landmarks(image):
49
+ """Get landmarks from PIL image"""
50
+ shape_predictor = dlib.shape_predictor(MODEL_PATH)
51
+
52
+ max_size = max(image.size)
53
+ reduction_scale = int(max_size/512)
54
+ if reduction_scale == 0:
55
+ reduction_scale = 1
56
+ downscaled = image.reduce(reduction_scale)
57
+ img = np.array(downscaled)
58
+ detections = detector(img, 0)
59
+
60
+ for detection in detections:
61
+ try:
62
+ face_landmarks = [(reduction_scale*item.x, reduction_scale*item.y) for item in shape_predictor(img, detection).parts()]
63
+ yield face_landmarks
64
+ except Exception as e:
65
+ print(e)
66
+
67
+
68
+ def image_align(src_img, face_landmarks, output_size=512, transform_size=2048, enable_padding=True, x_scale=1, y_scale=1, em_scale=0.1, alpha=False):
69
+ # Align function modified from ffhq-dataset
70
+ # See https://github.com/NVlabs/ffhq-dataset for license
71
+
72
+ lm = np.array(face_landmarks)
73
+ lm_eye_left = lm[2:3] # left-clockwise
74
+ lm_eye_right = lm[0:1] # left-clockwise
75
+
76
+ # Calculate auxiliary vectors.
77
+ eye_left = np.mean(lm_eye_left, axis=0)
78
+ eye_right = np.mean(lm_eye_right, axis=0)
79
+ eye_avg = (eye_left + eye_right) * 0.5
80
+ eye_to_eye = 0.71*(eye_right - eye_left)
81
+ mouth_avg = lm[4]
82
+ eye_to_mouth = 1.35*(mouth_avg - eye_avg)
83
+
84
+ # Choose oriented crop rectangle.
85
+ x = eye_to_eye.copy()
86
+ x /= np.hypot(*x)
87
+ x *= max(np.hypot(*eye_to_eye) * 2.0, np.hypot(*eye_to_mouth) * 1.8)
88
+ x *= x_scale
89
+ y = np.flipud(x) * [-y_scale, y_scale]
90
+ c = eye_avg + eye_to_mouth * em_scale
91
+ quad = np.stack([c - x - y, c - x + y, c + x + y, c + x - y])
92
+ quad_orig = quad.copy()
93
+ qsize = np.hypot(*x) * 2
94
+
95
+ img = src_img.convert('RGBA').convert('RGB')
96
+
97
+ # Shrink.
98
+ shrink = int(np.floor(qsize / output_size * 0.5))
99
+ if shrink > 1:
100
+ rsize = (int(np.rint(float(img.size[0]) / shrink)), int(np.rint(float(img.size[1]) / shrink)))
101
+ img = img.resize(rsize, Image.ANTIALIAS)
102
+ quad /= shrink
103
+ qsize /= shrink
104
+
105
+ # Crop.
106
+ border = max(int(np.rint(qsize * 0.1)), 3)
107
+ crop = (int(np.floor(min(quad[:,0]))), int(np.floor(min(quad[:,1]))), int(np.ceil(max(quad[:,0]))), int(np.ceil(max(quad[:,1]))))
108
+ crop = (max(crop[0] - border, 0), max(crop[1] - border, 0), min(crop[2] + border, img.size[0]), min(crop[3] + border, img.size[1]))
109
+ if crop[2] - crop[0] < img.size[0] or crop[3] - crop[1] < img.size[1]:
110
+ img = img.crop(crop)
111
+ quad -= crop[0:2]
112
+
113
+ # Pad.
114
+ pad = (int(np.floor(min(quad[:,0]))), int(np.floor(min(quad[:,1]))), int(np.ceil(max(quad[:,0]))), int(np.ceil(max(quad[:,1]))))
115
+ pad = (max(-pad[0] + border, 0), max(-pad[1] + border, 0), max(pad[2] - img.size[0] + border, 0), max(pad[3] - img.size[1] + border, 0))
116
+ if enable_padding and max(pad) > border - 4:
117
+ pad = np.maximum(pad, int(np.rint(qsize * 0.3)))
118
+ img = np.pad(np.float32(img), ((pad[1], pad[3]), (pad[0], pad[2]), (0, 0)), 'reflect')
119
+ h, w, _ = img.shape
120
+ y, x, _ = np.ogrid[:h, :w, :1]
121
+ mask = np.maximum(1.0 - np.minimum(np.float32(x) / pad[0], np.float32(w-1-x) / pad[2]), 1.0 - np.minimum(np.float32(y) / pad[1], np.float32(h-1-y) / pad[3]))
122
+ blur = qsize * 0.02
123
+ img += (gaussian_filter(img, [blur, blur, 0]) - img) * np.clip(mask * 3.0 + 1.0, 0.0, 1.0)
124
+ img += (np.median(img, axis=(0,1)) - img) * np.clip(mask, 0.0, 1.0)
125
+ img = np.uint8(np.clip(np.rint(img), 0, 255))
126
+ if alpha:
127
+ mask = 1-np.clip(3.0 * mask, 0.0, 1.0)
128
+ mask = np.uint8(np.clip(np.rint(mask*255), 0, 255))
129
+ img = np.concatenate((img, mask), axis=2)
130
+ img = Image.fromarray(img, 'RGBA')
131
+ else:
132
+ img = Image.fromarray(img, 'RGB')
133
+ quad += pad[:2]
134
+
135
+ # Transform.
136
+ img = img.transform((transform_size, transform_size), Image.QUAD, (quad + 0.5).flatten(), Image.BILINEAR)
137
+ if output_size < transform_size:
138
+ img = img.resize((output_size, output_size), Image.ANTIALIAS)
139
+
140
+ return img, quad_orig
ffhq2pumpkinheads_pix2pixHD_e03_itr4500-simp.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3dc9bc6902dbc85c7168adf07ed999d4f5f7638bafa23166a5b8c17166851a9
3
+ size 729804826
requirements.txt ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ numpy
2
+ opencv-python-headless
3
+ Pillow
4
+ scikit-image
5
+ onnxruntime
6
+ scipy
7
+ cmake
shape_predictor_5_face_landmarks.dat ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4b1e9804792707d3a405c2c16a80a20269e6675021f64a41d30fffafbc41888
3
+ size 9150489