I try to use yolov3-mobilenet-v2 on my jetson nano and when I create the execution context it give me this error:
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-7-5e5d94a8baad> in __init__(self, model, input_shape, category_num, cuda_ctx)
26 # 使用create_execution_context创建可推理上下文
---> 27 self.context = self.engine.create_execution_context()
28 self.inputs, self.outputs, self.bindings, self.stream = allocate_buffers(self.engine)
AttributeError: 'NoneType' object has no attribute 'create_execution_context'
The above exception was the direct cause of the following exception:
RuntimeError Traceback (most recent call last)
<ipython-input-15-7e90f933ff27> in <module>
1 #12
2 from IPython.display import Image
----> 3 main_one()
4 Image("result.jpg")
<ipython-input-12-0b6b13d5f07a> in main_one()
6 cls_dict = get_cls_dict("yolo_mobilenet_v2_traffic".split('_')[-1])
7 model_name ="yolo_mobilenet_v2_traffic"
----> 8 trtYOLO = TrtYOLO(model_name, INPUT_HW)
9 vis = BBoxVisualization(cls_dict)
10 print("start detection!")
<ipython-input-7-5e5d94a8baad> in __init__(self, model, input_shape, category_num, cuda_ctx)
29 # 返回报错信息
30 except Exception as e:
---> 31 raise RuntimeError('fail to allocate CUDA resources') from e
32 finally:
33 if self.cuda_ctx:
RuntimeError: fail to allocate CUDA resources
---------------------------------------------------------------------------
for the environment:
1. Jetpack 4.4
2. Python 3.6.9
3. TensorRT 7.1
4. CUDA 10.2
5. Numpy 1.16.1
6. Opencv 4.1.1
Since I use jupyter, I cannot copy all the code directly here, here is the main function for inference：
class TrtYOLO(object):
"""TrtYOLO class encapsulates things needed to run TRT YOLO."""
def _load_engine(self):
# 打开yolo/model_name.bin文件，以二进制只读格式
TRTbin = 'yolo/%s.bin' % self.model
with open(TRTbin, 'rb') as f, trt.Runtime(self.trt_logger) as runtime:
return runtime.deserialize_cuda_engine(f.read())
def __init__(self, model, input_shape, category_num=80, cuda_ctx=None):
"""Initialize TensorRT plugins, engine and conetxt."""
self.model = model
self.input_shape = input_shape
self.category_num = category_num
self.cuda_ctx = cuda_ctx
if self.cuda_ctx:
self.cuda_ctx.push()
self.inference_fn = do_inference if trt.__version__[0] < '7' else do_inference_v2
self.trt_logger = trt.Logger(trt.Logger.INFO)
self.engine = self._load_engine()
try:
self.context = self.engine.create_execution_context()
self.inputs, self.outputs, self.bindings, self.stream = allocate_buffers(self.engine)
except Exception as e:
raise RuntimeError('fail to allocate CUDA resources') from e
finally:
if self.cuda_ctx:
self.cuda_ctx.pop()
def __del__(self):
"""Free CUDA memories."""
del self.outputs
del self.inputs
del self.stream
def detect(self, img, conf_th=0.3):
"""Detect objects in the input image."""
img_resized = _preprocess_yolo(img, self.input_shape)
# Set host input to the image. The do_inference() function
# will copy the input to the GPU before executing.
self.inputs[0].host = np.ascontiguousarray(img_resized)
if self.cuda_ctx:
self.cuda_ctx.push()
trt_outputs = self.inference_fn(
context=self.context,
bindings=self.bindings,
inputs=self.inputs,
outputs=self.outputs,
stream=self.stream)
if self.cuda_ctx:
self.cuda_ctx.pop()
boxes, scores, classes = _postprocess_yolo(trt_outputs, img.shape[1], img.shape[0], conf_th)
# clip x1, y1, x2, y2 within original image
boxes[:, [0, 2]] = np.clip(boxes[:, [0, 2]], 0, img.shape[1]-1)
boxes[:, [1, 3]] = np.clip(boxes[:, [1, 3]], 0, img.shape[0]-1)
return boxes, scores, classes
Here are some documents about the model I use and other related documents:
https://drive.google.com/file/d/1Iw-4sRZNFUwH-gXM-L_vwgD8-jcHln/view?usp=sharing