To anyone having trouble converting tensorflow models with dilated convolutions, make sure you are using the correct dilated convolution implementation. Tensorflow has tf.nn.atrous_conv2d | TensorFlow v2.10.0 and tf.nn.conv2d | TensorFlow v2.10.0 The first one will wrap normal(non dilated) conv2d ops with space_to_batch and batch_to_space as a way of implementing the dilation. The second link is the op that can use the cudnn implementations for dilated convolutions and also makes for a graph that converts nicely with the uff parser. That being said, I’m still experiencing some issues on tx2 with dilated convolutions. Even more strange, the same frozen tensorflow graph converted/running on my 1080ti works perfectly…
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| dilation layer could not work in TensorRT 3.0 RC of Jetson TX2 | 10 | 1509 | October 18, 2021 | |
| Tensorflow's dilated convolutions with UFF parser not supported | 2 | 1765 | October 18, 2021 | |
| TensorRT dilated convolution | 0 | 959 | August 11, 2017 | |
| about Space to Batch and Batch to Space operation | 7 | 1445 | October 18, 2021 | |
| Depthwise Convolution and space to Batch conversion | 0 | 773 | June 7, 2019 | |
| TensorFlow OP SpaceToBatchND Does Not Work Correctly on TX2 | 9 | 936 | October 18, 2021 | |
| convert tensorflow model use tf.nn.convolution have dilation_rate!=1 to uff out error | 3 | 999 | June 7, 2019 | |
| some layer params not work when using tensorRT2 to accelerate caffe | 2 | 772 | October 18, 2021 | |
| SpaceToDepth supported in tensorRT? | 6 | 1315 | July 22, 2019 | |
| Uffparser error for deconvolution in conv2d_transpose | 5 | 2153 | September 7, 2018 |