メニュー
Normalize raw stroke coordinates to 0-255 range, resample at uniform arc-length spacing, then apply Ramer-Douglas-Peucker simplification
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Memory-efficient Keras generator that streams sharded CSV files in chunks, renders strokes to images on-the-fly, and yields batches for training on datasets too large for memory
Stack 1D convolutions for local feature extraction before bidirectional LSTMs to classify variable-length stroke sequences into hundreds of doodle categories
Partition a large dataset into N balanced shards using integer key modulo arithmetic for reproducible, class-interleaved splits across CSV files
Render stroke sequences to grayscale images with temporal intensity encoding where earlier strokes are brighter and later strokes fade to encode drawing order
Score each face in a video as anomalous by computing L2 distance from the embedding centroid of all faces, then convert to probability via logistic function
Extract per-frame CNN features then classify the temporal sequence with stacked GRU layers and a boolean mask for variable-length video inputs
| name | cv-stroke-normalize-simplify-pipeline |
| description | Normalize raw stroke coordinates to 0-255 range, resample at uniform arc-length spacing, then apply Ramer-Douglas-Peucker simplification |
Raw hand-drawn strokes have arbitrary coordinate ranges and irregular point spacing. A three-step pipeline — normalize to [0, 255], resample at uniform arc-length intervals, then simplify with Ramer-Douglas-Peucker — produces clean, compact stroke representations. This reduces point count by 3-5x while preserving shape fidelity, improving both storage efficiency and model performance.
import numpy as np
import math
def resample(x, y, spacing=1.0):
output = []
px, py = x[0], y[0]
cumlen, offset = 0, 0
for i in range(1, len(x)):
dx, dy = x[i] - px, y[i] - py
seg_len = math.sqrt(dx*dx + dy*dy)
cumlen += seg_len
while offset < cumlen:
t = (offset - (cumlen - seg_len)) / seg_len
output.append((px + t*dx, py + t*dy))
offset += spacing
px, py = x[i], y[i]
output.append((x[-1], y[-1]))
return np.array(output)
def normalize_strokes(strokes):
all_pts = np.concatenate([np.array(s).T for s in strokes])
mn, mx = all_pts.min(axis=0), all_pts.max(axis=0)
rng = max(mx - mn)
result = []
for s in strokes:
pts = np.array(s, dtype=float).T
pts = (pts - mn) / rng * 255
resampled = resample(pts[:, 0], pts[:, 1], spacing=1.0)
result.append(np.round(resampled).astype(np.uint8).T.tolist())
return result
(pt - min) / max_range * 255