ai.onnxruntime.OrtException: Error code - ORT_INVALID_ARGUMENT

[umi-soft]

Novice Question

Train with yolov8，option batch value 64，run your code has errors

OnnxTensor.createTensor ERROR

ai.onnxruntime.OrtException: Error code - ORT_INVALID_ARGUMENT - message: Got invalid dimensions for input: images for the following indices
 index: 0 Got: 1 Expected: 64
 Please fix either the inputs or the model.

if i set option batch value 16 and retrain，your code can run success。 I guess I need to make changes NUM_INPUT_ELEMENTS and INPUT_SHAPE，but I am clear about the specific meaning of the values and how to modify them.

[umi-soft]

The image size is also 640

[jhgan00]

Hello. Currently, my code is written without considering dynamic batch sizes. In other words, the code only works with a batch size of 1. Based on the error message, it seems that you exported the trained model to ONNX with a batch size of 64. If you could share the trained model or provide information about the input shape through apps like Netron, I would be able to assist you further.

[jhgan00]

• If you exported the trained model to ONNX with a batch size of 64, then during inference, you need to feed exactly 64 images to the model.
• If you want to use a variable number of input images instead of a fixed number, you should export the model to ONNX with dynamic batch size support.

In both cases, we need to modify the current code

[umi-soft]

thanks, i get it.