hkust-aaron's picture
Add annotated sample pack: quality, timeline, clip captions, keyframe masks (v1/v2), SAM 3.1 dense tracks
07f5340 verified
Raw History Blame Contribute Delete
2.22 kB
"""Preview sheets for the dense SAM 3.1 tracks.
Usage: python 09_dense_preview.py <video> <dense_tracks_dir>
Draws the masks of 24 evenly spaced frames (and, per clip, frames 0 / 125 / 249)
over the video and writes dense_tracks/previews/overview.jpg and
previews/clip_XX.jpg. Colour = class; text-prompt tracks are grey-blue.
"""
import json
import sys
from pathlib import Path
import cv2
import numpy as np
video, root = sys.argv[1], Path(sys.argv[2])
COL = {"white_insulated_jawed": (255, 200, 0), "dark_rod": (0, 165, 255), "fenestrated_grasper": (0, 255, 0),
"gauze": (255, 0, 255), "instrument_text_detected": (200, 160, 120)}
meta = {int(p.stem[5:]): json.load(open(p)) for p in sorted((root / "tracks").glob("clip_*.json"))}
cap = cv2.VideoCapture(video)
def overlay(f):
cap.set(cv2.CAP_PROP_POS_FRAMES, f)
ok, img = cap.read()
c = f // 250
inst = cv2.imread(str(root / "masks" / f"clip_{c:02d}" / f"{f:05d}.png"), 0)
for tid in np.unique(inst)[1:]:
m = (inst == tid).astype(np.uint8)
col = COL[meta[c]["tracks"][str(tid)]["category"]]
img[m == 1] = (0.55 * img[m == 1] + 0.45 * np.array(col)).astype(np.uint8)
cnts, _ = cv2.findContours(m, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cv2.drawContours(img, cnts, -1, col, 2)
ys, xs = np.nonzero(m)
cv2.putText(img, str(tid), (int(xs.mean()), int(ys.mean())), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (255, 255, 255), 2)
cv2.putText(img, f"frame {f}", (8, 24), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 255, 255), 2)
return cv2.resize(img, (360, 288))
def sheet(frames, cols):
tiles = [overlay(f) for f in frames]
while len(tiles) % cols:
tiles.append(np.zeros_like(tiles[0]))
return np.vstack([np.hstack(tiles[i:i + cols]) for i in range(0, len(tiles), cols)])
out = root / "previews"
out.mkdir(exist_ok=True)
cv2.imwrite(str(out / "overview.jpg"), sheet([int(x) for x in np.linspace(0, 17369, 24)], 6), [cv2.IMWRITE_JPEG_QUALITY, 85])
for c, m in meta.items():
s, n = m["start_frame"], m["num_frames"]
cv2.imwrite(str(out / f"clip_{c:02d}.jpg"), sheet([s, s + n // 2, s + n - 1], 3), [cv2.IMWRITE_JPEG_QUALITY, 85])
print("previews", len(meta) + 1)