TOKUNAGA Hiroyuki

Since we already have TensorView class, we can rewrite all parts which depends on MatrixView class with TensorView class.

Consider upgrading to TensorFlow 2.0

11

TensorFlow 2.0 was released. We should consider upgrading our TF dependency to 2.0. However, since this is large major version up, we still don't know how much source code needs...

[WIP] a new network GlazedYolo

4

This network is experimental, so far it could not run on FPGAs. ## Description This network brings several recent ideas to our YOLOv2 implementation. In short, GlazedYolo = YoloV2 +...

use byte arrays to remove inexact floating point number literals

2

## What this patch does to fix the issue. Conversion between floating point value :left_right_arrow: decimal string representation cause some small error. In this pull request, I removed floating point...

consider to remove matplotlib dependency

4

It might be possible to remove a dependency matplotlib. (Though I'm not sure...)

add new preprocessing: PerImageStandardization

4

Currently, we supports only `DivideBy255` preprocessor. However, internally, we've prepared `PerImageStandardization`. @yasumura-lm san mentioned that sometimes this preprocessing method is is very effective. Hence, we want to support `PerImageStandardization`.

Optimize Add implementation

1

According to the performance measurement of LmResnet for ImageNet on FPGA, there is some low hanging fruits, `Add` operator is now consuming over 20ms.