Does TAO Toolkit support object detection models with “String” datatype?
Could you elaborate more? Thanks.
It is a model trained with Google AutoML to detect labels in boxes (input data is the image, not video). The model has input and output nodes with “String” datatype.
Does TAO Toolkit support object detection models with “String” datatype nodes?
Any link for this kind of model? Is it similar to OCR model?
I will send it in private message
As synced offline, the user case is similar to optical characters detection. Currently, TAO does not support this yet.
There is no update from you for a period, assuming this is not an issue anymore.
Hence we are closing this topic. If need further support, please open a new one.
TensorRT support for String datatype has been filed as a feature request here:
[Bug/Feature Request] -Native TensorRT to support “string” datatype for CV models · Issue #2026 · NVIDIA/TensorRT · GitHub