python : Parsing json file into list of dictionaries - json

Question

I have the following json file annotations
and here is a screenshot form it.tree structure of the json file
I want to parse it and extract the following info
here is a link which I take this screenshot form it Standard Dataset Dicts
I tried to use this code which is not working as expected.
def get_buildings_dicts(img_dir):
json_file = os.path.join(img_dir, "annotations.json")
with open(json_file) as f:
imgs_anns = json.load(f)
dataset_dicts = []
for idx, v in enumerate(imgs_anns):
record = {}
filename = os.path.join(img_dir, v["imagePath"])
height, width = cv2.imread(filename).shape[:2]
record["file_name"] = filename
record["image_id"] = idx
record["height"] = height
record["width"] = width
annos = v["shapes"][idx]
objs = []
for anno in annos:
# assert not anno["region_attributes"]
anno = anno["shape_type"]
px = anno["points"][0]
py = anno["points"][1]
poly = [(x + 0.5, y + 0.5) for x, y in zip(px, py)]
poly = [p for x in poly for p in x]
obj = {
"bbox": [np.min(px), np.min(py), np.max(px), np.max(py)],
"bbox_mode": BoxMode.XYXY_ABS,
"segmentation": [poly],
"category_id": 0,
}
objs.append(obj)
record["annotations"] = objs
dataset_dicts.append(record)
return dataset_dicts
here is an expected output of the final dict items:
{
"file_name": "balloon/train/34020010494_e5cb88e1c4_k.jpg",
"image_id": 0,
"height": 1536,
"width": 2048,
"annotations": [
{
"bbox": [994, 619, 1445, 1166],
"bbox_mode": <BoxMode.XYXY_ABS: 0>,
"segmentation": [[1020.5, 963.5, 1000.5, 899.5, 994.5, 841.5, 1003.5, 787.5, 1023.5, 738.5, 1050.5, 700.5, 1089.5, 663.5, 1134.5, 638.5, 1190.5, 621.5, 1265.5, 619.5, 1321.5, 643.5, 1361.5, 672.5, 1403.5, 720.5, 1428.5, 765.5, 1442.5, 800.5, 1445.5, 860.5, 1441.5, 896.5, 1427.5, 942.5, 1400.5, 990.5, 1361.5, 1035.5, 1316.5, 1079.5, 1269.5, 1112.5, 1228.5, 1129.5, 1198.5, 1134.5, 1207.5, 1144.5, 1210.5, 1153.5, 1190.5, 1166.5, 1177.5, 1166.5, 1172.5, 1150.5, 1174.5, 1136.5, 1170.5, 1129.5, 1153.5, 1122.5, 1127.5, 1112.5, 1104.5, 1084.5, 1061.5, 1037.5, 1032.5, 989.5, 1020.5, 963.5]],
"category_id": 0
}
]
}

Answer 1

I think the only tricky part is dealing with the nested lists but a handful of coprehensions can probably make life easier for us.
Try:
import json
new_images = []
with open("merged_file.json", "r") as file_in:
for index, image in enumerate( json.load(file_in)):
#height, width = cv2.imread(filename).shape[:2]
height, width = 100, 100
new_images.append({
"image_id": index,
"filename": image["imagePath"],
"height": height,
"width": width,
"annotations": [
{
"category_id": 0,
#"bbox_mode": BoxMode.XYXY_ABS,
"bbox_mode": 0,
"bbox": [
min(x for x,y in shape["points"]),
min(y for x,y in shape["points"]),
max(x for x,y in shape["points"]),
max(y for x,y in shape["points"])
],
"segmentation": [coord for point in shape["points"] for coord in point]
}
for shape in image["shapes"]
],
})
print(json.dumps(new_images, indent=2))