我想用TensorRT来加速YOLO v2模型，但是YOLO v2模型中包含有BatchNorm层，程序会在ICudaEngine* engine = builder-&g

如何处理网络模型中的BatchNorm层？ about tensorrt_tutorial HOT 4 CLOSED

litleo commented on June 25, 2024

如何处理网络模型中的BatchNorm层？

from tensorrt_tutorial.

Comments (4)

rencheng123456 commented on June 25, 2024

你好，我最近也在研究使用TensorRT来加速YOLO v2模型，想问下，但不知道如何着手。希望能给予我一些指导

from tensorrt_tutorial.

jsmlay commented on June 25, 2024

TensorRT 3.0 Have YOLO plugins headers. Plz double check before usage.
Find it in NvInferPlugin.h in /usr/src/TensorRT/include.
It need Jetpack 3.2 support.
When deploy it, you need to write plugin, plz refer to FasterRCNN sample.
Sorry for using English because i'm on TX2.

from tensorrt_tutorial.

LitLeo commented on June 25, 2024

抱歉最近没怎么关注git。BN的实现可以用scale层代替，直接将BN层的公式转换成scale层公式就可以了。我之前的实现代码如下，仅供参考。

                float *scale_values = new float[l_ptr->out_c];
                float *shift_values = new float[l_ptr->out_c];
                for (int i = 0; i < l_ptr->out_c; ++i)
                {
                    scale_values[i] = 1 / (sqrt(l_ptr->rolling_variance[i] + .00001f)) * l_ptr->scales[i];
                    shift_values[i] = -l_ptr->rolling_mean[i] / (sqrt(l_ptr->rolling_variance[i] + .00001f)) * l_ptr->scales[i] + l_ptr->biases[i];
                }
                Weights scale {DataType::kFLOAT, scale_values, l_ptr->out_c};
                Weights shift {DataType::kFLOAT, shift_values, l_ptr->out_c};
                Weights power {DataType::kFLOAT, nullptr, 0};

                auto scale_1 = network->addScale(*inputs_v[0], ScaleMode::kCHANNEL, shift, scale, power);
                assert(scale_1 != nullptr);
                inputs_v[0] = scale_1->getOutput(0);

from tensorrt_tutorial.