How to effectively read ".yml" files as structures?

247 vues (au cours des 30 derniers jours)
cui,xingxing
cui,xingxing le 8 Déc 2020
Modifié(e) : cui,xingxing le 22 Nov 2022
The ".yml" file is also a storage structure similar to the ".xml" file. How do I read the content inside and save it as a struct type structure or dictionary type(introduced in R2022b)? Similar to the new function readstruct function in 2020b.
Similarly, whether format files such as'.json' can be read in matlab with a unified function, such as the readstruct function, the specified'FileType' can be {'.xml','.json','.yml' ,'.xml'} One or more than one kind of'.xml', although there are functions such as jsondecode and xmlread.
My ".yml" file is similar to the following format(in attachments):
#Config File example
save_dir: workspace/nanodet_m
model:
arch:
name: GFL
backbone:
name: ShuffleNetV2
model_size: 1.0x
out_stages: [2,3,4]
activation: LeakyReLU
fpn:
name: PAN
in_channels: [116, 232, 464]
out_channels: 96
start_level: 0
num_outs: 3
head:
name: NanoDetHead
num_classes: 80
input_channel: 96
feat_channels: 96
stacked_convs: 2
share_cls_reg: True
octave_base_scale: 5
scales_per_octave: 1
strides: [8, 16, 32]
reg_max: 7
norm_cfg:
type: BN
loss:
loss_qfl:
name: QualityFocalLoss
use_sigmoid: True
beta: 2.0
loss_weight: 1.0
loss_dfl:
name: DistributionFocalLoss
loss_weight: 0.25
loss_bbox:
name: GIoULoss
loss_weight: 2.0
data:
train:
name: coco
img_path: coco/train2017
ann_path: coco/annotations/instances_train2017.json
input_size: [320,320] #[w,h]
keep_ratio: True
  2 commentaires
cui,xingxing
cui,xingxing le 2 Fév 2021
I found a third-party solution, but I need to compile opencv with mex, cv.FileStorage reads and writes various formats above, but I still feel that it is not what I really want. I very much hope that the official version will strengthen the readstruct function in the future!

Connectez-vous pour commenter.

Réponse acceptée

Martin Koch
Martin Koch le 17 Fév 2022
Hi, I wrote a wrapper for SnakeYAML with which you can load and dump YAML:
Requires R2019b or newer.
  12 commentaires
cui,xingxing
cui,xingxing le 15 Juin 2022
Modifié(e) : cui,xingxing le 15 Juin 2022
@Walter Roberson,Yes, this package does not currently suitable for the yml import and export attached in my comment.
But on the other hand, the attached yml format is produced by the opencv library, not our common yml format, is unique to opencv, so to achieve the same function in matlab, can only package the opencv filestorage class interface for matlab to use, as you said in your comments the year before the third-party cvyamlparser library
Martin Koch
Martin Koch le 19 Juin 2022
Hi, here is a copy of my answer from the GitHub issue:
there are two problems:
  • The underlying SnakeYAML is written for YAML 1.1 for which the first line %YAML:1.0 is not valid (%YAML 1.0 without : would be acceptable). I will update the documentation to mention YAML 1.1.
  • The custom opencv types are not supported.
You can build a workaround like this:
  • Read file into string with fileread
  • Delete all text occurrences of %YAML:1.0 and !!opencv-matrix (e.g. with strrep)
  • Parse modified string with yaml.load
  • Manually convert the content of the data field to a 2D array (e.g. with reshape)
To write YAML files in the same format you have to reverse the above process. However, it will not be formatted such that every row is in a new line.

Connectez-vous pour commenter.

Plus de réponses (1)

Walter Roberson
Walter Roberson le 2 Fév 2021
  2 commentaires
cui,xingxing
cui,xingxing le 2 Fév 2021
Thank you, this may be a better solution before the official unified read interface function (readstruct).
Razvan
Razvan le 20 Oct 2022
Just for others to know. The function from https://github.com/llerussell/ReadYAML does not properly deals with multi level Yaml files.

Connectez-vous pour commenter.

Produits

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by