Labelme 轉coco
Labelme 標註圖像生成的json格式
{
"version": "4.2.10",
"flags": {},
"shapes": [# 每個對象的形狀
{ # 第一個對象
"label": "malignant",
"line_color": null,
"fill_color": null,
"points": [# 邊緣是由點構成,將這些點連在一起就是對象的邊緣多邊形
[
371, # 第一個點 x 座標
257 # 第一個點 y 座標
],
...
[
412,
255
]
],
"shape_type": "polygon" # 形狀類型:多邊形
},
{
"label": "malignant", # 第一個對象的標籤
"line_color": null,
"fill_color": null,
"points": [# 第二個對象
[
522,
274
],
...
[
561,
303
]
],
"shape_type": "polygon"
},
"imagePath": "/your/images/10_image_1.png"", # 原始圖片的路徑
"imageData":"something too long ", #原圖像數據 通過該字段可以解析出原圖像數據
"imageHeight": 2134,
"imageWidth": 1508
}
coco標準數據集格式
COCO通過大量使用Amazon Mechanical Turk來收集數據。COCO數據集現在有3種標註類型:object instances(目標實例), object keypoints(目標上的關鍵點), and image captions(看圖說話),使用JSON文件存儲
基本的JSON結構體類型
這3種類型共享下面所列的基本類型,包括image、categories、annotation類型。
- Images類型
"images": [
{
"height": 768,
"width": 1024,
"id": 1, #圖片id
"file_name": "000002.jpg"
}
]
- categories類型
"categories": [
{
"supercategory": "Cancer", #父類
"id": 1, #標籤類別id,0表示背景
"name": "benign" #子類
},
{
"supercategory": "Cancer",
"id": 2,
"name": "malignant"
}
],
- annotations類型
"annotations": [
{
"segmentation": [#座標點的座標值
[
418,
256,
391,
293,
406,
323,
432,
340,
452,
329,
458,
311,
458,
286,
455,
277,
439,
264,
418,
293,
391,
256
]
],
"iscrowd": 0, #單個的對象(iscrowd=0)可能需要多個polygon來表示
"image_id": 1, #和image的id保持一致
"bbox": [ #標註的邊框值 bbox是將segmentation包起來的水平矩形
391.0,
256.0,
67.0,
84.0
],
"area": 5628.0, #標註的邊框面積
"category_id": 1, #所屬類別id
"id": 1 #標註邊框的id : 1,2,3...,n
}
]
coco數據集主要是images,categories,annotations(數據格式都是list)
coco是把所有圖片的信息都整合在一起了,成了一個dict
images:主要有height,width,id,file_name#id是圖片id
categories:主要有supercategory,id,name#id是category的id
annotations:主要有segmentation,iscrowd,image_id,bbox,area,category_id,id
要注意在annotations裏id之間的關係
coco的格式是
{
images:[{height,width,id,file_name}, {height,width,id,file_name}…]
categories:[{supercategory,id,name},{supercategory,id,name}…]
annotations:[{segmentation,iscrowd,image_id,bbox,area,category_id,id},
{segmentation,iscrowd,image_id,bbox,area,category_id,id}]
}
轉爲coco格式
import os, sys
import json
import cv2
import base64
import numpy as np
import glob
import PIL
from PIL import Image, ImageDraw
def base64_cv2(base64_str):
"""
base64轉cv2
"""
imgString = base64.b64decode(base64_str)
nparr = np.fromstring(imgString, np.uint8)
image = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
return image
class MyEncoder(json.JSONEncoder):
def default(self, obj):
if isinstance(obj, np.integer):
return int(obj)
elif isinstance(obj, np.floating):
return float(obj)
elif isinstance(obj, np.ndarray):
return obj.tolist()
else:
return super(MyEncoder, self).default(obj)
class labelme2coco(object):
def __init__(self, labelme_json=[], save_json_path="./tran.json"):
"""
:param labelme_json: 所有labelme的json文件路徑組成的列表
:param save_json_path: json保存位置
"""
self.labelme_json = labelme_json
self.save_json_path = save_json_path
self.images = []
self.categories = []
self.annotations = []
# self.data_coco = {}
self.label = []
self.annID = 1
self.height = 0
self.width = 0
self.save_json()
def data_transfer(self):
for num, json_file in enumerate(self.labelme_json):
with open(json_file, "r") as fp:
data = json.load(fp) # 加載json文件
self.images.append(self.image(data, num))
for shapes in data["shapes"]:
label = shapes["label"]
if label not in self.label:
self.categories.append(self.categorie(label))
self.label.append(label)
points = shapes["points"] # 這裏的point是用rectangle標註得到的,只有兩個點,需要轉成四個點
# points.append([points[0][0],points[1][1]])
# points.append([points[1][0],points[0][1]])
self.annotations.append(self.annotation(points, label, num))
self.annID += 1
def image(self, data, num):
image = {}
img = base64_cv2(data["imageData"]) # 解析原圖片數據
filename = data["imagePath"].split('\\')[-1]
# filename 根據labelme json的imagePath來獲取
image["file_name"] = filename
cv2.imwrite(f"train/{filename}", img)
height, width = img.shape[:2]
image["height"] = height
image["width"] = width
image["id"] = num + 1
self.height = height
self.width = width
return image
def categorie(self, label):
categorie = {}
categorie["supercategory"] = "Cancer"
categorie["id"] = len(self.label) + 1 # 0 默認爲背景
categorie["name"] = label
return categorie
def annotation(self, points, label, num):
annotation = {}
annotation["segmentation"] = [list(np.asarray(points).flatten())]
annotation["iscrowd"] = 0
annotation["image_id"] = num + 1
annotation["bbox"] = list(map(float, self.getbbox(points)))
annotation["area"] = annotation["bbox"][2] * annotation["bbox"][3]
# annotation['category_id'] = self.getcatid(label)
annotation["category_id"] = self.getcatid(label) # 注意,源代碼默認爲1
annotation["id"] = self.annID
return annotation
def getcatid(self, label):
for categorie in self.categories:
if label == categorie["name"]:
return categorie["id"]
return 1
def getbbox(self, points):
polygons = points
mask = self.polygons_to_mask([self.height, self.width], polygons)
return self.mask2box(mask)
def mask2box(self, mask):
"""從mask反算出其邊框
mask:[h,w] 0、1組成的圖片
1對應對象,只需計算1對應的行列號(左上角行列號,右下角行列號,就可以算出其邊框)
"""
# np.where(mask==1)
index = np.argwhere(mask == 1)
rows = index[:, 0]
clos = index[:, 1]
# 解析左上角行列號
left_top_r = np.min(rows) # y
left_top_c = np.min(clos) # x
# 解析右下角行列號
right_bottom_r = np.max(rows)
right_bottom_c = np.max(clos)
return [
left_top_c,
left_top_r,
right_bottom_c - left_top_c,
right_bottom_r - left_top_r,
] # [x1,y1,w,h] 對應COCO的bbox格式
def polygons_to_mask(self, img_shape, polygons):
mask = np.zeros(img_shape, dtype=np.uint8)
mask = PIL.Image.fromarray(mask)
xy = list(map(tuple, polygons))
ImageDraw.Draw(mask).polygon(xy=xy, outline=1, fill=1)
mask = np.array(mask, dtype=bool)
return mask
def data2coco(self):
data_coco = {}
data_coco["images"] = self.images
data_coco["categories"] = self.categories
data_coco["annotations"] = self.annotations
return data_coco
def save_json(self):
self.data_transfer()
self.data_coco = self.data2coco()
# 保存json文件
json.dump(
self.data_coco, open(self.save_json_path, "w"), indent=4, cls=MyEncoder
) # indent=4 更加美觀顯示
if __name__ == '__main__':
labelme_json = glob.glob("./labelme_json/*.json")
labelme2coco(labelme_json, "train.json")
print(f"*************** labelme2coco done ***************")