sam3-demo / app.py
Yanlin Zhang
use sam3 pipeline
ccd869c
"""
Vehicle trajectory extractor powered by SAM3.
The app takes an aerial video, segments small and large vehicles frame-by-frame
with text prompts (`small-vehicle`, `large-vehicle`), and draws their
trajectories on top of the footage.
"""
from __future__ import annotations
import math
import os
import tempfile
import uuid
from dataclasses import dataclass
from typing import Dict, List, Sequence, Tuple
import cv2
import gradio as gr
import numpy as np
from PIL import Image
import torch
from transformers import pipeline
# -----------------------------------------------------------------------------
# Configuration
# -----------------------------------------------------------------------------
MODEL_ID = "facebook/sam3"
TEXT_PROMPTS = ["small-vehicle", "large-vehicle"]
MIN_MASK_PIXELS = 150 # filter spurious detections
MAX_TRACK_GAP = 3 # frames
DEFAULT_FRAME_STRIDE = 5
MAX_PROCESSED_FRAMES = 720
CLASS_COLORS: Dict[str, Tuple[int, int, int]] = {
"small-vehicle": (20, 148, 245), # RGB
"large-vehicle": (255, 120, 30),
}
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
# -----------------------------------------------------------------------------
# Model + processor
# -----------------------------------------------------------------------------
# Use pipeline as shown in Hugging Face guidance
# Then extract model and processor for text-prompt support
mask_pipe = pipeline("mask-generation", model=MODEL_ID, device=0 if DEVICE == "cuda" else -1)
# Extract model and processor from pipeline for direct text prompt usage
model = mask_pipe.model
processor = mask_pipe.feature_extractor if hasattr(mask_pipe, 'feature_extractor') else mask_pipe.image_processor
# -----------------------------------------------------------------------------
# Tracking utilities
# -----------------------------------------------------------------------------
@dataclass
class Track:
track_id: int
label: str
points: List[Tuple[int, float, float]]
last_frame: int
score: float | None
def _extract_detections(frame_rgb: np.ndarray) -> List[Dict]:
pil_image = Image.fromarray(frame_rgb)
detections: List[Dict] = []
for label in TEXT_PROMPTS:
# Use processor and model directly with text prompt
try:
inputs = processor(images=pil_image, text=label, return_tensors="pt")
inputs = {
k: (v.to(DEVICE) if isinstance(v, torch.Tensor) else v)
for k, v in inputs.items()
}
with torch.inference_mode():
outputs = model(**inputs)
# Extract masks from outputs - SAM3 outputs structure may vary
if hasattr(outputs, "pred_masks"):
masks = outputs.pred_masks
elif hasattr(outputs, "masks"):
masks = outputs.masks
elif isinstance(outputs, dict):
masks = outputs.get("pred_masks") or outputs.get("masks")
else:
masks = outputs
if masks is None:
continue
# Handle different mask formats
if isinstance(masks, torch.Tensor):
if masks.ndim == 4: # [batch, num_masks, H, W]
masks = masks[0] # Remove batch dimension
elif masks.ndim == 3: # [num_masks, H, W]
pass
else:
continue
for mask_tensor in masks:
mask_np = mask_tensor.squeeze().detach().cpu().numpy()
if mask_np.ndim == 3:
mask_np = mask_np[0]
binary_mask = mask_np > 0.5
area = int(binary_mask.sum())
if area < MIN_MASK_PIXELS:
continue
ys, xs = np.nonzero(binary_mask)
if len(xs) == 0:
continue
centroid = (float(xs.mean()), float(ys.mean()))
detections.append(
{
"label": label,
"mask": binary_mask,
"score": None,
"centroid": centroid,
"area": area,
}
)
except Exception as e:
# Fallback to pipeline if direct access fails
try:
results = mask_pipe(pil_image)
if not isinstance(results, list):
results = [results]
for result in results:
if isinstance(result, dict):
mask = result.get("mask")
score = result.get("score")
else:
mask = result
score = None
if isinstance(mask, Image.Image):
mask_np = np.array(mask.convert("L"))
elif isinstance(mask, torch.Tensor):
mask_np = mask.squeeze().detach().cpu().numpy()
elif isinstance(mask, np.ndarray):
mask_np = mask
else:
continue
if mask_np.ndim == 3:
mask_np = mask_np[:, :, 0] if mask_np.shape[2] == 1 else mask_np.max(axis=2)
if mask_np.max() > 1.0:
mask_np = mask_np / 255.0
binary_mask = mask_np > 0.5
area = int(binary_mask.sum())
if area < MIN_MASK_PIXELS:
continue
ys, xs = np.nonzero(binary_mask)
if len(xs) == 0:
continue
centroid = (float(xs.mean()), float(ys.mean()))
detections.append(
{
"label": label,
"mask": binary_mask,
"score": float(score) if score is not None else None,
"centroid": centroid,
"area": area,
}
)
except Exception as e2:
raise gr.Error(f"Both direct model access and pipeline failed: {e2}")
return detections
def _update_tracks(
tracks: List[Track],
detections: Sequence[Dict],
frame_idx: int,
max_distance: float,
) -> None:
for detection in detections:
centroid = np.array(detection["centroid"])
best_track = None
best_distance = math.inf
for track in tracks:
if track.label != detection["label"]:
continue
if frame_idx - track.last_frame > MAX_TRACK_GAP:
continue
prev_point = np.array(track.points[-1][1:])
dist = np.linalg.norm(centroid - prev_point)
if dist < best_distance and dist <= max_distance:
best_distance = dist
best_track = track
if best_track:
best_track.points.append((frame_idx, *detection["centroid"]))
best_track.last_frame = frame_idx
best_track.score = detection["score"]
else:
new_track = Track(
track_id=len(tracks) + 1,
label=detection["label"],
points=[(frame_idx, *detection["centroid"])],
last_frame=frame_idx,
score=detection["score"],
)
tracks.append(new_track)
def _blend_mask(frame: np.ndarray, mask: np.ndarray, color: Tuple[int, int, int], alpha: float = 0.45):
overlay = frame.copy()
overlay[mask] = (1 - alpha) * overlay[mask] + alpha * np.array(color, dtype=np.float32)
return overlay
def _draw_annotations(
frame_rgb: np.ndarray,
detections: Sequence[Dict],
tracks: Sequence[Track],
frame_idx: int,
):
annotated = frame_rgb.astype(np.float32)
for det in detections:
color_rgb = CLASS_COLORS.get(det["label"], (255, 255, 255))
color_bgr = tuple(int(c) for c in reversed(color_rgb))
annotated = _blend_mask(annotated, det["mask"], color_rgb)
cx, cy = det["centroid"]
cv2.circle(annotated, (int(cx), int(cy)), 4, color_bgr, -1)
cv2.putText(
annotated,
det["label"],
(int(cx) + 4, int(cy) - 4),
cv2.FONT_HERSHEY_SIMPLEX,
0.4,
color_bgr,
1,
cv2.LINE_AA,
)
for track in tracks:
if len(track.points) < 2:
continue
if track.points[-1][0] < frame_idx - MAX_TRACK_GAP:
continue
color_rgb = CLASS_COLORS.get(track.label, (255, 255, 255))
color_bgr = tuple(int(c) for c in reversed(color_rgb))
pts = [
(int(x), int(y))
for (f_idx, x, y) in track.points
if f_idx <= frame_idx
]
for i in range(1, len(pts)):
cv2.line(annotated, pts[i - 1], pts[i], color_bgr, 2, cv2.LINE_AA)
cv2.circle(annotated, pts[-1], 5, color_bgr, -1)
return np.clip(annotated, 0, 255).astype(np.uint8)
def _summarize_tracks(tracks: Sequence[Track]) -> List[Dict]:
summary = []
for track in tracks:
if len(track.points) < 2:
continue
distances = []
for (prev_frame, x1, y1), (curr_frame, x2, y2) in zip(track.points, track.points[1:]):
distances.append(math.sqrt((x2 - x1) ** 2 + (y2 - y1) ** 2))
summary.append(
{
"track_id": track.track_id,
"label": track.label,
"frames": len(track.points),
"start_frame": track.points[0][0],
"end_frame": track.points[-1][0],
"path_px": round(float(sum(distances)), 2),
}
)
return summary
# -----------------------------------------------------------------------------
# Video processing
# -----------------------------------------------------------------------------
def analyze_video(
video_path: str,
frame_stride: int = DEFAULT_FRAME_STRIDE,
max_frames: int = MAX_PROCESSED_FRAMES,
resize_long_edge: int = 1280,
) -> Tuple[str, List[Dict]]:
if not video_path:
raise gr.Error("Please upload an aerial video (MP4, MOV, ...).")
capture = cv2.VideoCapture(video_path)
if not capture.isOpened():
raise gr.Error("Unable to read the uploaded video.")
fps = capture.get(cv2.CAP_PROP_FPS) or 15
width = int(capture.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(capture.get(cv2.CAP_PROP_FRAME_HEIGHT))
diag = math.sqrt(width**2 + height**2)
max_assign_distance = 0.04 * diag
processed_frames = []
tracks: List[Track] = []
frame_index = 0
processed_count = 0
while processed_count < max_frames:
ret, frame_bgr = capture.read()
if not ret:
break
if frame_index % frame_stride != 0:
frame_index += 1
continue
frame_rgb = cv2.cvtColor(frame_bgr, cv2.COLOR_BGR2RGB)
frame_rgb = _resize_long_edge(frame_rgb, resize_long_edge)
detections = _extract_detections(frame_rgb)
_update_tracks(tracks, detections, frame_index, max_assign_distance)
annotated = _draw_annotations(frame_rgb, detections, tracks, frame_index)
processed_frames.append(cv2.cvtColor(annotated, cv2.COLOR_RGB2BGR))
processed_count += 1
frame_index += 1
capture.release()
if not processed_frames:
raise gr.Error("No frames were processed. Try lowering the stride or uploading a different video.")
output_path = _write_video(processed_frames, fps / max(frame_stride, 1))
summary = _summarize_tracks(tracks)
return output_path, summary
def _resize_long_edge(frame_rgb: np.ndarray, target_long_edge: int) -> np.ndarray:
h, w, _ = frame_rgb.shape
long_edge = max(h, w)
if long_edge <= target_long_edge:
return frame_rgb
scale = target_long_edge / long_edge
new_size = (int(w * scale), int(h * scale))
resized = cv2.resize(frame_rgb, new_size, interpolation=cv2.INTER_AREA)
return resized
def _write_video(frames: Sequence[np.ndarray], fps: float) -> str:
height, width, _ = frames[0].shape
tmp_path = os.path.join(tempfile.gettempdir(), f"sam3-trajectories-{uuid.uuid4().hex}.mp4")
fourcc = cv2.VideoWriter_fourcc(*"mp4v")
writer = cv2.VideoWriter(tmp_path, fourcc, max(fps, 1.0), (width, height))
for frame in frames:
writer.write(frame)
writer.release()
return tmp_path
# -----------------------------------------------------------------------------
# Gradio UI
# -----------------------------------------------------------------------------
with gr.Blocks(title="SAM3 Vehicle Trajectories") as demo:
gr.Markdown(
"""
### SAM3 for Vehicle Trajectories
1. Upload an aerial surveillance video.
2. The app prompts SAM3 with `small-vehicle` and `large-vehicle`.
3. Segmentations are linked across frames to render motion trails.
"""
)
with gr.Row():
video_input = gr.Video(label="Aerial video (MP4/MOV)")
controls = gr.Column()
with controls:
stride_slider = gr.Slider(
label="Frame stride",
minimum=1,
maximum=12,
value=DEFAULT_FRAME_STRIDE,
step=1,
info="Process one frame every N frames",
)
max_frames_slider = gr.Slider(
label="Max frames to process",
minimum=30,
maximum=1000,
value=MAX_PROCESSED_FRAMES,
step=10,
)
resize_slider = gr.Slider(
label="Resize longest edge (px)",
minimum=640,
maximum=1920,
value=1280,
step=40,
)
output_video = gr.Video(label="Overlay with trajectories")
track_table = gr.Dataframe(
headers=["track_id", "label", "frames", "start_frame", "end_frame", "path_px"],
datatype=["number", "str", "number", "number", "number", "number"],
wrap=True,
label="Track summary",
)
run_button = gr.Button("Extract trajectories", variant="primary")
run_button.click(
fn=analyze_video,
inputs=[video_input, stride_slider, max_frames_slider, resize_slider],
outputs=[output_video, track_table],
api_name="analyze",
)
if __name__ == "__main__":
demo.launch()