how to change the keypoints in a model.
I wanted to use something like centernet_resnet50_v2_512x512_kpts_coco17_tpu-8
Which does OBJ detection and keypoints.
I have learned how to change the classes, I know how to add the keypoints on the label maps and the pictures themselves.
But how to I add it to the premade TS model above.
I don’t want to reinvent the wheel here just be able to alter for my own needs.
That model does the face and body, I want to do birds.
is there anyone who knows how to change the key points?
Bhack
June 22, 2021, 3:21pm
#2
There was already a thread at:
opened 07:39PM - 29 Mar 21 UTC
type:docs
models:research:odapi
# Prerequisites
Please answer the following question for yourself before subm… itting an issue.
- [*] I checked to make sure that this issue has not been filed already.
## 1. The entire URL of the documentation with the issue
https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/configuring_jobs.md
## 2. Describe the issue
I am trying to train a CenterNet model using custom keypoints.
However, I can't find any documentation or example on how to configure the CenterNet pipeline file.
In https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf2_detection_zoo.md, the examples are just about a human pose and there is not even an explanation on how to set the **keypoint_estimation_task**, or what the **value** parameter of **keypoint_label_to_std** means or how to calculate it. The same for **keypoint_label_to_sigmas**.
Example of **centernet_mobilenetv2fpn_512x512_coco17_kpts.tar.gz**:
> keypoint_label_map_path: "PATH_TO_BE_CONFIGURED/label_map.txt"
> keypoint_estimation_task {
> task_name: "human_pose"
> task_loss_weight: 1.0
> loss {
> localization_loss {
> l1_localization_loss {
> }
> }
> classification_loss {
> penalty_reduced_logistic_focal_loss {
> alpha: 2.0
> beta: 4.0
> }
> }
> }
> keypoint_class_name: "/m/01g317"
> keypoint_label_to_std {
> key: "left_ankle"
> value: 0.89
> }
> keypoint_label_to_std {
> key: "left_ear"
> value: 0.35
> }
> keypoint_label_to_std {
> key: "left_elbow"
> value: 0.72
> }
> ....
> keypoint_regression_loss_weight: 0.1
> keypoint_heatmap_loss_weight: 1.0
> keypoint_offset_loss_weight: 1.0
> offset_peak_radius: 3
> per_keypoint_offset: true
> }
1 Like