pythonでマスク処理した画像を3次元にする方法

下のように画像をマスク処理しました

python
1import cv2
2im=dataset.load_image(1)
3
4im_gray = 0.299 * im[:, :, 0] + 0.587 * im[:, :, 1] + 0.114 * im[:, :, 2]
5
6print(im.dtype)
7print(im_gray.dtype)
8# uint8
9# float64
10
11print(im.shape)
12print(im_gray.shape)
13# (512, 512, 3)
14# (512, 512)

このマスク処理した画像に整数のラベル番号を加え、3次元にする方法はないでしょうか？
label番号=51
だとすると**shape＝(512, 512)の画像に加えてshape=(512, 512, 51)**としたいのですが。ご教授お願いします

python
1# このようにしたい
2
3label=51
4print(add_label_img.shape)
5# (512, 512, 3)

hayataka2049

2019/05/25 21:45

shape=(512, 512, 51)にしたとして、この配列の中身はどうなっているのでしょうか。

行動規範の内容に同意します

回答1件

自己解決

python
1def load_mask(self, image_id):
2        """Load instance masks for the given image.
3        Different datasets use different ways to store masks. This
4        function converts the different mask format to one format
5        in the form of a bitmap [height, width, instances].
6        Returns:
7        masks: A bool array of shape [height, width, instance count] with
8            one mask per instance.
9        class_ids: a 1D array of class IDs of the instance masks.
10        """
11        # If not a COCO image, delegate to parent class.
12        image_info = self.image_info[image_id]
13        if image_info["source"] != "coco":
14            return super(CocoDataset, self).load_mask(image_id)
15
16        instance_masks = []
17        class_ids = []
18        annotations = self.image_info[image_id]["annotations"]
19        # Build mask of shape [height, width, instance_count] and list
20        # of class IDs that correspond to each channel of the mask.
21        for annotation in annotations:
22            class_id = self.map_source_class_id(
23                "coco.{}".format(annotation['category_id']))
24            if class_id:
25                m = self.annToMask(annotation, image_info["height"],
26                                   image_info["width"])
27                # Some objects are so small that they're less than 1 pixel area
28                # and end up rounded out. Skip those objects.
29                if m.max() < 1:
30                    continue
31                # Is it a crowd? If so, use a negative class ID.
32                if annotation['iscrowd']:
33                    # Use negative class ID for crowds
34                    class_id *= -1
35                    # For crowd masks, annToMask() sometimes returns a mask
36                    # smaller than the given dimensions. If so, resize it.
37                    if m.shape[0] != image_info["height"] or m.shape[1] != image_info["width"]:
38                        m = np.ones([image_info["height"], image_info["width"]], dtype=bool)
39                instance_masks.append(m)
40                class_ids.append(class_id)
41
42        # Pack instance masks into an array
43        if class_ids:
44            mask = np.stack(instance_masks, axis=2).astype(np.bool)
45            class_ids = np.array(class_ids, dtype=np.int32)
46            return mask, class_ids
47        else:
48            # Call super class to return an empty mask
49            return super(CocoDataset, self).load_mask(image_id)