-
Notifications
You must be signed in to change notification settings - Fork 0
/
model_config_seg_kp.yaml
38 lines (31 loc) · 1.21 KB
/
model_config_seg_kp.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
# Specify things which we want to train multitask netowrk on. Base netowrk perorms object Detection using FasterRCN
# valid types are regression, classification and segmetation and keypoints (all the image points can be specified in keypoints).
# Make sure the dataloder loads all the label names specified by names below
# Background plus Required classes
numclasses: 2
classes: ["cucumber"]
# Provide label name, classes and file location with labels for segmentation
segmentation:
name: ["masks"]
# classes for which the above segmentation task is for
classes: [["cucumber"]]
file: None
# Provide label name, class, number of classes and file location with labels for classification
classification:
name: None
# For which class object the classification is required
classes: None
numclasses: None
file: None
# Provide label name, class and file location with labels for regression
regression:
name: [None]
# For which class the Regression is required
class: [None]
file: None
# Provide label name, classes, num of keypoints and file location with labels for keypoint detection
keypoints:
name: ["head","tail"]
classes: ["cucumber"]
num: 2
file: None