-
Notifications
You must be signed in to change notification settings - Fork 288
Open
Labels
feature_requestNew feature or requestNew feature or request
Description
Preliminary Checks
- This issue is not a duplicate. Before opening a new issue, please search existing issues.
- This issue is not a question, bug report, or anything other than a feature request directly related to this project.
Proposal
As evident in the ZED SDK custom object detection sample, the onnx dynamic input shape can be set to a non-square resolution.
detection_parameters.custom_onnx_dynamic_input_shape = sl::Resolution(320, 320);
But in the ros2 wrapper it is set to a single integer value, which I assume default constructs a square sl::Resolution with width and height equal to mYoloOnnxSize.
od_p.custom_onnx_dynamic_input_shape = mYoloOnnxSize
I propose allowing the object_detection.custom_onnx_input_size ros2 parameter to be an array or adding object_detection.custom_onnx_input_width and object_detection.custom_onnx_input_height, such that non-square input tensors can be used.
Use-Case
Running YOLO inference on a non-square input tensor.
Anything else?
No response
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
feature_requestNew feature or requestNew feature or request