Spaces:
Sleeping
Sleeping
File size: 7,724 Bytes
f3841e1 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 | """
Visualization utilities for facial landmark and symmetry analysis.
"""
from __future__ import annotations
import cv2
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
from matplotlib.figure import Figure
from src.landmarks.detector import LandmarkResult, LANDMARK_GROUPS, BILATERAL_PAIRS
from src.symmetry.analyzer import SymmetryResult
# Color palette (BGR for cv2, RGB for matplotlib)
COLORS_BGR = {
"left_eye": (86, 180, 233),
"right_eye": (86, 180, 233),
"left_eyebrow": (230, 159, 0),
"right_eyebrow": (230, 159, 0),
"nose_tip": (0, 158, 115),
"nose_bridge": (0, 158, 115),
"upper_lip": (204, 121, 167),
"lower_lip": (204, 121, 167),
"jawline": (160, 160, 160),
"midline": (255, 255, 255),
}
def draw_landmarks(
image: np.ndarray,
result: LandmarkResult,
radius: int = 2,
draw_connections: bool = True,
alpha: float = 0.7,
) -> np.ndarray:
"""Draw 468 facial landmarks on the image.
Args:
image: RGB uint8 image.
result: LandmarkResult from FaceLandmarkDetector.
radius: Dot radius in pixels.
draw_connections: Draw group contour lines.
alpha: Blend factor for overlay.
Returns:
RGB uint8 annotated image.
"""
overlay = image.copy()
bgr = cv2.cvtColor(overlay, cv2.COLOR_RGB2BGR)
# Draw all landmarks as small dots (gray)
for pt in result.landmarks:
cv2.circle(bgr, (int(pt[0]), int(pt[1])), radius, (180, 180, 180), -1)
# Draw group landmarks with distinct colors
for group_name, pts in result.groups.items():
color = COLORS_BGR.get(group_name, (255, 255, 255))
rgb_color = (color[2], color[1], color[0]) # BGR → RGB for drawing on RGB img
for pt in pts:
cv2.circle(bgr, (int(pt[0]), int(pt[1])), radius + 1, color, -1)
if draw_connections and len(pts) > 1:
for i in range(len(pts) - 1):
p1 = (int(pts[i][0]), int(pts[i][1]))
p2 = (int(pts[i+1][0]), int(pts[i+1][1]))
cv2.line(bgr, p1, p2, color, 1, cv2.LINE_AA)
# Midline
mx = int(result.midline_x)
h = image.shape[0]
cv2.line(bgr, (mx, 0), (mx, h), (200, 200, 50), 1, cv2.LINE_AA)
result_rgb = cv2.cvtColor(bgr, cv2.COLOR_BGR2RGB)
return cv2.addWeighted(image, 1 - alpha, result_rgb, alpha, 0)
def draw_symmetry_overlay(
image: np.ndarray,
lm_result: LandmarkResult,
sym_result: SymmetryResult,
show_pairs: bool = True,
) -> np.ndarray:
"""Draw bilateral landmark pairs color-coded by asymmetry magnitude.
Green = low asymmetry, red = high asymmetry (relative to median).
Args:
image: RGB uint8 image.
lm_result: LandmarkResult.
sym_result: SymmetryResult.
show_pairs: Draw connecting lines between bilateral pairs.
Returns:
Annotated RGB image.
"""
bgr = cv2.cvtColor(image.copy(), cv2.COLOR_RGB2BGR)
pts = lm_result.landmarks
midline_x = lm_result.midline_x
distances = sym_result.pair_distances_px
# Normalize distances for color mapping
d_min, d_max = distances.min(), max(distances.max(), 1.0)
for i, (left_idx, right_idx) in enumerate(BILATERAL_PAIRS):
left_pt = pts[left_idx].astype(int)
right_pt = pts[right_idx].astype(int)
# Color: green (symmetric) → red (asymmetric)
t = (distances[i] - d_min) / (d_max - d_min)
color = (int(t * 255), int((1 - t) * 200), 30) # BGR
cv2.circle(bgr, tuple(left_pt), 4, color, -1)
cv2.circle(bgr, tuple(right_pt), 4, color, -1)
if show_pairs:
cv2.line(bgr, tuple(left_pt), tuple(right_pt), color, 1, cv2.LINE_AA)
# Midline
cv2.line(bgr, (int(midline_x), 0), (int(midline_x), image.shape[0]),
(80, 200, 200), 1, cv2.LINE_AA)
return cv2.cvtColor(bgr, cv2.COLOR_BGR2RGB)
def draw_bilateral_comparison(
image: np.ndarray,
lm_result: LandmarkResult,
) -> Figure:
"""Create a side-by-side left/right half comparison figure.
Flips the left half and places it next to the right half to visually
demonstrate what a perfectly symmetric version would look like.
Returns:
Matplotlib Figure (3 panels: original | left-mirrored | right-mirrored)
"""
h, w = image.shape[:2]
mx = int(lm_result.midline_x)
# Left half, mirrored (reflects left face to fill right side)
left_half = image[:, :mx]
left_mirror = np.fliplr(left_half)
# Pad or crop to match right half width
right_w = w - mx
if left_mirror.shape[1] >= right_w:
left_mirror = left_mirror[:, :right_w]
else:
pad = right_w - left_mirror.shape[1]
left_mirror = np.pad(left_mirror, ((0,0),(0,pad),(0,0)))
left_full = np.concatenate([np.fliplr(left_mirror), left_mirror], axis=1)
left_full = left_full[:, :w]
# Right half, mirrored
right_half = image[:, mx:]
right_mirror = np.fliplr(right_half)
left_w = mx
if right_mirror.shape[1] >= left_w:
right_mirror = right_mirror[:, :left_w]
else:
pad = left_w - right_mirror.shape[1]
right_mirror = np.pad(right_mirror, ((0,0),(0,pad),(0,0)))
right_full = np.concatenate([right_mirror, np.fliplr(right_mirror)], axis=1)
right_full = right_full[:, :w]
fig, axes = plt.subplots(1, 3, figsize=(15, 5))
axes[0].imshow(image); axes[0].set_title('Original', fontsize=12)
axes[1].imshow(left_full); axes[1].set_title('Left×2 (mirrored)', fontsize=12)
axes[2].imshow(right_full); axes[2].set_title('Right×2 (mirrored)', fontsize=12)
for ax in axes:
ax.axis('off')
plt.suptitle('Bilateral Symmetry Comparison', fontsize=14, fontweight='bold')
plt.tight_layout()
return fig
def plot_asymmetry_radar(sym_result: SymmetryResult) -> Figure:
"""Radar / spider chart of per-region symmetry scores."""
regions = list(sym_result.region_scores.keys())
scores = [sym_result.region_scores[r] for r in regions]
n = len(regions)
angles = np.linspace(0, 2 * np.pi, n, endpoint=False).tolist()
angles += angles[:1]
scores += scores[:1]
fig, ax = plt.subplots(figsize=(6, 6), subplot_kw=dict(polar=True))
ax.plot(angles, scores, 'o-', linewidth=2, color='royalblue')
ax.fill(angles, scores, alpha=0.25, color='royalblue')
ax.set_xticks(angles[:-1])
ax.set_xticklabels(regions, fontsize=11)
ax.set_ylim(0, 1)
ax.set_yticks([0.2, 0.4, 0.6, 0.8, 1.0])
ax.set_yticklabels(['0.2', '0.4', '0.6', '0.8', '1.0'], fontsize=8)
ax.set_title('Per-Region Symmetry Scores', fontsize=13, pad=20)
ax.grid(True, alpha=0.3)
return fig
def plot_asymmetry_bar(sym_result: SymmetryResult) -> Figure:
"""Horizontal bar chart of bilateral pair asymmetries (% IOD)."""
iod = sym_result.inter_ocular_distance
pairs_pct = sym_result.pair_distances_px / iod * 100
labels = sym_result.pair_labels
colors = ['#e74c3c' if v > 5 else '#2ecc71' for v in pairs_pct]
fig, ax = plt.subplots(figsize=(9, max(4, len(labels) * 0.5)))
y = np.arange(len(labels))
ax.barh(y, pairs_pct, color=colors, edgecolor='white', linewidth=0.5)
ax.set_yticks(y)
ax.set_yticklabels(labels, fontsize=10)
ax.set_xlabel('Asymmetry (% of inter-ocular distance)', fontsize=11)
ax.set_title('Bilateral Pair Asymmetry', fontsize=13)
ax.axvline(5, color='orange', linestyle='--', lw=1.5, label='5% IOD threshold')
ax.legend(fontsize=9)
ax.grid(axis='x', alpha=0.3)
plt.tight_layout()
return fig
|