Can`t use model 'ssd_mobilenet_v2_coco.engine'

environment version:

ubuntu      18.04 aarch64
jetpack      jetson-nano-jp451-sd-card-image
jetbot        v0.4.3
tensorflow     2.4.0+nv21.3
torch         1.6.0
torchvision        0.7.0a0+78ed10c
setuptools      49.6.0

model version:

ssd_mobilenet_v2_coco.engine v0.4 (latest)


from jetbot import ObjectDetector
from jetbot import Camera
model = ObjectDetector(‘ssd_mobilenet_v2_coco.engine’)
camera = Camera.instance(width=300, height=300)
detections = model.value(camera.value)

error information:

[TensorRT] ERROR: coreReadArchive.cpp (38) - Serialization Error in verifyHeader: 0 (Version tag does not match)
[TensorRT] ERROR: INVALID_STATE: std::exception
[TensorRT] ERROR: INVALID_CONFIG: Deserialize the cuda engine failed.
Traceback (most recent call last):
File “”, line 4, in
model = ObjectDetector(‘ssd_mobilenet_v2_coco.engine’)
File “/usr/local/lib/python3.6/dist-packages/jetbot-0.4.3-py3.6.egg/jetbot/”, line 29, in init
output_names=[TRT_OUTPUT_NAME, TRT_OUTPUT_NAME + ‘_1’])
File “/usr/local/lib/python3.6/dist-packages/jetbot-0.4.3-py3.6.egg/jetbot/”, line 59, in init
self.context = self.engine.create_execution_context()
AttributeError: ‘NoneType’ object has no attribute ‘create_execution_context’

If anyone can help me, I would be very grateful

This looks like a Jetson issue. moving to Jetson Nano forum for resolution.

1 Like

Jetson Nano based Jetbot issue it is.

1 Like


The error indicates the engine is compiled by a different TensorRT version.

Please noted that since the engine file doesn’t support portability.
You will need to re-create the model on your environment directly.


1 Like

Thank for your reply.

First thanks for your advice. Could you tell me where I can find the issue?

Thank for your advice.

Did you follow the official Jetbot page?

How about try the old version from NVIDIA-AI-IOT firstly?

I try jetbot 0.3.0 and use SSD model 0.3, it work.

Please note, the latest object detection engine targets Jetson Nano running software bundled with JetPack 4.3. This engine not be compatible with other versions of JetPack. For this reason, we recommend starting from the JetPack 4.3 JetBot SD Card image, or using the JetPack 4.3 base SD card image if you want to run the object detection example.


I think some version issue you have between JetPack and model file.

1 Like

Yes, this should be the core of the problem. Thank you MtHiker for your reply.

Please refer this:

1 Like

Yes, I get the answer in this page.