The pointer should be 64 bits on my virtual machine. After correcting it, I
have deployed the Faster R-CNN on VTA. Actually solving this problem, the
Faster R-CNN can be supported by VTA.
```
auto data_ptr_tmp = static_cast(input->data);
auto data_ptr = reinterpret_cast(*data_ptr_tmp);
au
At present, there is few problem with quantization. The following work is to
modify the graph pack function to transform most convolutions into NCHW1n16c to
get accelerating. I need to add some op names to complete AST traverse in graph
pack function. If there is a mistake, please correct me.
tvm.lower python api, you need to give the schedule and input/output symbol.
```
print(tvm.lower(s, [data, valid_count, out], name="test_nms"))
```
---
[Visit
Topic](https://discuss.tvm.ai/t/vta-a-workaround-for-deploying-faster-r-cnn-on-target-ext-dev-vta-and-arm-cpu/6516/5)
to respond.
@thierry I have accelerated the 42-layer convolution on vta. I choose
faster_rcnn_resnet50_v1b_voc mxnet model which has 56-layer convolution. I am
going to work with my partner to do some optimization @c
---
[Vi
https://drive.google.com/open?id=1io_uQjG9am5mYbFQ-c7h9nH07fYLmVnq
Here is my project including .so files. You can unzip it and run the
fasterRCNN_vta.py directly with fsim for vta. I didn't make a git commit
because there are some compatibility problems in my code. You can use git
status to
