alpha

https://github.com/ch

Results 4 comments of


                                            alpha

I have convert nano.trt success,but do infer can't get right result

yes

I have convert nano.trt success,but do infer can't get right result

my tensorrt=7.0.0.11,export onnx I have change opset=10.0

I have convert nano.trt success,but do infer can't get right result

I convert FP32 can get success right result,but convert fp16 can't get success right result

rk3399使用ACL存在内存泄露？,开启ACL内存一直涨,最终会killed；关闭ACL使用cpu推理内存保持不变

操作:graph是个全局变量，先加载模型初始化graph，然后每次推理只调用run_graph操作