UFF SSD sample resize

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Hello. I have managed to run the sampleUffSSD both with Float and Int8. But the input is fixed to 300x300 as I understood. Is there any way to change the input size? I have tried changing the INPUT_H and INPUT_W in the C++ files as well as in config.py, but it throws the following error while parsing the model:

ERROR: UFFParser: Parser error: BoxPredictor_0/Reshape: Reshape: Volume mismatch
ERROR: sample_uff_ssd: Fail to parse
sample_uff_ssd: sampleUffSSD.cpp:667: int main(int, char**): Assertion `tmpEngine != nullptr' failed.
2

Hello,

The TF SSD network the sample uses currently is trained for images of size 300x300. So all the current network parameters are set accordingly. To use input images of different size, a resize function will be required that resizes the image to 300x300

We plan to have a python uff ssd sample soon which will work with images of different sizes automatically. We have received some requests of this and are discussing internally.
Sorry again that we cannot share more information about further release here.
Please pay attention to our announcement for the information.

3

Thanks