mirror of
https://github.com/azaion/annotations.git
synced 2026-04-22 09:46:30 +00:00
Alex Bezdieniezhnykh
e798af470b
automate tensorrt model conversion in case of no existing one for user's gpu
38 lines
932 B
Cython
38 lines
932 B
Cython
from typing import List, Tuple
|
|
import numpy as np
|
|
|
|
|
|
cdef class InferenceEngine:
|
|
cdef public int batch_size
|
|
cdef tuple get_input_shape(self)
|
|
cdef int get_batch_size(self)
|
|
cdef get_class_names(self)
|
|
cpdef run(self, input_data)
|
|
|
|
cdef class OnnxEngine(InferenceEngine):
|
|
cdef object session
|
|
cdef list model_inputs
|
|
cdef str input_name
|
|
cdef object input_shape
|
|
cdef object class_names
|
|
|
|
cdef class TensorRTEngine(InferenceEngine):
|
|
cdef object stream
|
|
cdef object context
|
|
cdef str input_name
|
|
cdef str output_name
|
|
cdef object d_input
|
|
cdef object d_output
|
|
cdef object input_shape
|
|
cdef object output_shape
|
|
cdef object h_output
|
|
cdef object class_names
|
|
|
|
@staticmethod
|
|
cdef bytes convert_from_onnx(bytes onnx_model)
|
|
|
|
@staticmethod
|
|
cdef unsigned long long get_gpu_memory_bytes(device_id=?)
|
|
|
|
@staticmethod
|
|
cdef str get_engine_filename(device_id=?) |