Issue with Batch Prediction in Vertex AI Image Segmentation using DeepLabv3+

[LiamC]

I have encountered an issue while using a DeepLabv3+ model from Model Garden in Google Vertex AI's model registry. When attempting batch prediction, the process appears successful in the batch predictions page. However, reviewing the logs, I only find error messages, and when multiple inputs are provided to the model, it produces fewer predicted items than the total number of items, probably due to the errors. The complete error log can be found here. I would appreciate any insights or assistance in resolving this issue.