segment-anything 根据box分割
发布时间
阅读量：
阅读量
该文本介绍了使用SAM（Segment Anything Model）进行图像分割的技术。代码展示了如何通过SAM模型对图像进行单个box和批量预测，并通过OpenCV和PyTorch库实现。具体包括：
使用SAM模型对单个区域进行分割，并显示分割后的掩膜和矩形框
批量处理多个区域的分割预测，并对每个分割结果进行显示
使用PyTorch加载SAM模型，设置设备并读取图像
通过SAM模型的预测函数生成多通道掩膜，并结合输入框进行显示
整理了根据box分割box内容，如下图，可以用来分割标注：
目录
一个box，交互分割代码：
多个box批量预测：
一个box，交互分割代码：
import time
using_colab = False
import numpy as np
import torch
# import matplotlib.pyplot as plt
import cv2
def show_points(coords, labels, ax, marker_size=375):
pos_points = coords[labels == 1]
neg_points = coords[labels == 0]
ax.scatter(pos_points[:, 0], pos_points[:, 1], color='green', marker='*', s=marker_size, edgecolor='white',
linewidth=1.25)
ax.scatter(neg_points[:, 0], neg_points[:, 1], color='red', marker='*', s=marker_size, edgecolor='white',
linewidth=1.25)
if using_colab:
import torch
import torchvision
print("PyTorch version:", torch.__version__)
print("Torchvision version:", torchvision.__version__)
print("CUDA is available:", torch.cuda.is_available())
import sys
def show_mask(mask, ax, random_color=False):
if random_color:
color = np.concatenate([np.random.random(3), np.array([0.6])], axis=0)
else:
color = np.array([30 / 255, 144 / 255, 255 / 255, 0.6])
h, w = mask.shape[-2:]
mask_image = mask.reshape(h, w, 1) * color.reshape(1, 1, -1)
cv2.imshow("one", mask_image)
cv2.waitKey(0)
ax.imshow(mask_image)
if __name__ == '__main__':
import sys
sys.path.append("..")
from segment_anything import sam_model_registry, SamPredictor
sam_checkpoint = "sam_vit_h_4b8939.pth"
model_type = "vit_h"
sam_checkpoint = "sam_vit_l_0b3195.pth"
model_type = "vit_l"
device = "cuda:0"
image_o = cv2.imread(r'F:\3d\Python-3D-Rasterizer-main\pyrender-main\res\50\00004_img.jpg')
roi = cv2.selectROI(windowName="roi", img=image_o, showCrosshair=True, fromCenter=False)
x, y, w, h = roi
input_box = np.array([x, y, x + w, y + h])
image = cv2.cvtColor(image_o, cv2.COLOR_BGR2RGB)
sam = sam_model_registry[model_type](checkpoint=sam_checkpoint)
sam.to(device=device)
predictor = SamPredictor(sam)
for i in range(1):
start = time.time()
predictor.set_image(image)
masks, _, _ = predictor.predict(
point_coords=None,
point_labels=None,
box=input_box[None, :],
multimask_output=False,
)
print('time', (time.time() - start))
# plt.figure(figsize=(10, 10))
# plt.imshow(image)
# show_mask(masks[0], plt.gca())
random_color=False
mask=masks[0]
if random_color:
color = np.concatenate([np.random.random(3), np.array([0.6])], axis=0)
else:
color = np.array([30 / 255, 144 / 255, 255 / 255, 0.6])
h, w = mask.shape[-2:]
mask_image = mask.reshape(h, w, 1) * color.reshape(1, 1, -1)
x0, y0 = input_box[0], input_box[1]
w, h = input_box[2] - input_box[0], input_box[3] - input_box[1]
cv2.rectangle(image_o, (x0, y0), (input_box[2], input_box[3]), (0, 255, 0), 2)
cv2.imshow("img", image_o)
cv2.imshow("one", mask_image)
cv2.waitKey(0)
# show_box(input_box, plt.gca())
# plt.axis('off')
# plt.show()多个box批量预测：
import time
using_colab = False
import numpy as np
import torch
# import matplotlib.pyplot as plt
import cv2
def show_points(coords, labels, ax, marker_size=375):
pos_points = coords[labels == 1]
neg_points = coords[labels == 0]
ax.scatter(pos_points[:, 0], pos_points[:, 1], color='green', marker='*', s=marker_size, edgecolor='white',
linewidth=1.25)
ax.scatter(neg_points[:, 0], neg_points[:, 1], color='red', marker='*', s=marker_size, edgecolor='white',
linewidth=1.25)
if using_colab:
import torch
import torchvision
print("PyTorch version:", torch.__version__)
print("Torchvision version:", torchvision.__version__)
print("CUDA is available:", torch.cuda.is_available())
import sys
def show_mask(mask, ax, random_color=False):
if random_color:
color = np.concatenate([np.random.random(3), np.array([0.6])], axis=0)
else:
color = np.array([30 / 255, 144 / 255, 255 / 255, 0.6])
h, w = mask.shape[-2:]
mask_image = mask.reshape(h, w, 1) * color.reshape(1, 1, -1)
cv2.imshow("one", mask_image)
cv2.waitKey(0)
ax.imshow(mask_image)
if __name__ == '__main__':
import sys
sys.path.append("..")
from segment_anything import sam_model_registry, SamPredictor
sam_checkpoint = "sam_vit_h_4b8939.pth"
model_type = "vit_h"
sam_checkpoint = "sam_vit_l_0b3195.pth"
model_type = "vit_l"
device = "cuda:0"
image_o = cv2.imread(r'F:\3d\img.jpg')
image = cv2.cvtColor(image_o, cv2.COLOR_BGR2RGB)
sam = sam_model_registry[model_type](checkpoint=sam_checkpoint)
sam.to(device=device)
predictor = SamPredictor(sam)
input_boxes = torch.tensor([
[839, 28, 953, 149],
[1377, 2, 1519, 287],
], device=predictor.device)
transformed_boxes = predictor.transform.apply_boxes_torch(input_boxes, image.shape[:2])
predictor.set_image(image)
masks, _, _ = predictor.predict_torch(
point_coords=None,
point_labels=None,
boxes=transformed_boxes,
multimask_output=False,
)
print(masks.shape)
for index,mask in enumerate(masks):
input_box=input_boxes[index]
random_color = False
if random_color:
color = np.concatenate([np.random.random(3), np.array([0.6])], axis=0)
else:
color = np.array([30 / 255, 144 / 255, 255 / 255, 0.6])
h, w = mask.shape[-2:]
mask_image = mask.cpu().numpy().reshape(h, w, 1) * color.reshape(1, 1, -1)
x0, y0 = int(input_box[0]), int(input_box[1])
w, h = input_box[2] - input_box[0], input_box[3] - input_box[1]
cv2.rectangle(image_o, (x0, y0), (int(input_box[2]),int(input_box[3])), (0, 255, 0), 2)
cv2.imshow("img", image_o)
cv2.imshow("one", mask_image)
cv2.waitKey(0)
# plt.figure(figsize=(10, 10))
# plt.imshow(image)
# show_mask(masks[0], plt.gca())
# show_box(input_box, plt.gca())
# plt.axis('off')
# plt.show()全部评论 (0)
还没有任何评论哟~