search

ShuangXieIrene / ssds.pytorch

Repository for Single Shot MultiBox Detector and its variants, implemented with pytorch, python3.
MIT License
568 stars 165 forks source link

run in jetson nano raise error... RuntimeError:cuda runtime error(7):too many resources requested for launch at..........how to solve it #77

Closed DENESTY closed 4 years ago

QZ-cmd commented 4 years ago

jetson nano can‘t train model

DENESTY commented 4 years ago

@QZ-cmd The model was trained on the x86 computer，Errors occur in inference on jetson nano

foreverYoungGitHub commented 4 years ago

I do not have the jetson nano to test the model, but I can use jetson tx2 to run the trt model converted from ssds.pytorch. Did you try to convert the model to trt?

foreverYoungGitHub commented 4 years ago

@DENESTY Just tested the converted trt model on the Jetson tx2 today, and everything works fine. If you need I can share how to convert the model from ssds.pytorch to tensorrt.

Since the issues is not existed on the trt model on the Jetson tx2. I'll close the issue for now.