NvDsPreProcessTensorMeta

class pyds.NvDsPreProcessTensorMeta

Tensor meta containing prepared tensor and related info inside preprocess user meta which is attached at batch level.

Variables:

• raw_tensor_buffer – capsule, raw tensor buffer preprocessed for infer.

• max_elements_in_pool – int, size of raw tensor buffer.

• tensor_shape – list of int, raw tensor buffer shape.

• data_type – NvDsDataType, model datatype for which tensor prepared.

• tensor_name – string, to be same as model input layer name.

• gpu_id – int, gpu-id on which tensor prepared.

• private_data – capsule, pointer to buffer from tensor pool.

• meta_id – int, meta id for differentiating between multiple tensor meta from same gst buffer,for the case when sum of roi’s exceeds the batch size.

• maintain_aspect_ratio – bool, parameter to inform whether aspect ratio is maintained in the preprocess tensor.

cast(self: capsule) → pyds.NvDsPreProcessTensorMeta

Cast given object/data to NvDsPreProcessTensorMeta, call pyds.NvDsPreProcessTensorMeta.cast(data)