Spaces:
Sleeping
Sleeping
File size: 4,281 Bytes
f0573ed |
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 |
import cv2
import numpy as np
import torch
from metaseg import SamAutomaticMaskGenerator, sam_model_registry
from metaseg.utils.file import download_model
class SegAutoMaskGenerator:
def __init__(self):
self.model = None
self.device = "cuda" if torch.cuda.is_available() else "cpu"
def load_model(self, model_type):
if self.model is None:
model_path = download_model(model_type)
model = sam_model_registry[model_type](checkpoint=model_path)
model.to(device=self.device)
self.model = model
return self.model
def load_image(self, image_path):
image = cv2.imread(image_path)
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
return image
def load_video(self, video_path):
cap = cv2.VideoCapture(video_path)
frame_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
frame_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
fourcc = cv2.VideoWriter_fourcc(*"XVID")
fps = int(cap.get(cv2.CAP_PROP_FPS))
out = cv2.VideoWriter("output.mp4", fourcc, fps, (frame_width, frame_height))
return cap, out
def predict(self, frame, model_type, points_per_side, points_per_batch):
model = self.load_model(model_type)
mask_generator = SamAutomaticMaskGenerator(
model, points_per_side=points_per_side, points_per_batch=points_per_batch
)
masks = mask_generator.generate(frame)
return frame, masks
def save_image(self, source, model_type, points_per_side, points_per_batch):
read_image = self.load_image(source)
image, anns = self.predict(read_image, model_type, points_per_side, points_per_batch)
if len(anns) == 0:
return
sorted_anns = sorted(anns, key=(lambda x: x["area"]), reverse=True)
mask_image = np.zeros((anns[0]["segmentation"].shape[0], anns[0]["segmentation"].shape[1], 3), dtype=np.uint8)
colors = np.random.randint(0, 255, size=(256, 3), dtype=np.uint8)
for i, ann in enumerate(sorted_anns):
m = ann["segmentation"]
img = np.ones((m.shape[0], m.shape[1], 3), dtype=np.uint8)
color = colors[i % 256]
for i in range(3):
img[:, :, 0] = color[0]
img[:, :, 1] = color[1]
img[:, :, 2] = color[2]
img = cv2.bitwise_and(img, img, mask=m.astype(np.uint8))
img = cv2.addWeighted(img, 0.35, np.zeros_like(img), 0.65, 0)
mask_image = cv2.add(mask_image, img)
combined_mask = cv2.add(image, mask_image)
cv2.imwrite("output.jpg", combined_mask)
return "output.jpg"
def save_video(self, source, model_type, points_per_side, points_per_batch, min_area, max_area):
cap, out = self.load_video(source)
colors = np.random.randint(0, 255, size=(256, 3), dtype=np.uint8)
while True:
ret, frame = cap.read()
if not ret:
break
image, anns = self.predict(frame, model_type, points_per_side, points_per_batch)
if len(anns) == 0:
continue
sorted_anns = sorted(anns, key=(lambda x: x["area"]), reverse=True)
mask_image = np.zeros(
(anns[0]["segmentation"].shape[0], anns[0]["segmentation"].shape[1], 3), dtype=np.uint8
)
for i, ann in enumerate(sorted_anns):
if max_area > ann["area"] > min_area:
m = ann["segmentation"]
color = colors[i % 256] # Her nesne için farklı bir renk kullan
img = np.zeros((m.shape[0], m.shape[1], 3), dtype=np.uint8)
img[:, :, 0] = color[0]
img[:, :, 1] = color[1]
img[:, :, 2] = color[2]
img = cv2.bitwise_and(img, img, mask=m.astype(np.uint8))
img = cv2.addWeighted(img, 0.35, np.zeros_like(img), 0.65, 0)
mask_image = cv2.add(mask_image, img)
combined_mask = cv2.add(frame, mask_image)
out.write(combined_mask)
out.release()
cap.release()
cv2.destroyAllWindows()
return "output.mp4"
|