GitLab

* Added possibility to load PNG images from a data buffer.

* Fixed code not compiling with some versions of libpng that doesn't have const specifier.

* Used FileInfo struct as a single parameter for the read_image method.

error: calling a constexpr host function("log1p") from a device function("cuda_log1pexp") is not allowed. The experimental flag '--expt-relaxed-constexpr' can be used to allow this.

The error only happens with some versions of CUDA.

Previously they were restored to default values, which had the effect of negating the current setting of the calling code (whether set in the compiler options or by pragma previously)

Improve the error a user gets when trying to build the pyhton extention on windows but they didn't install visual C++.

paths instead of relying on CUDA_PATH to be set.

* * Corrected interpolate_bilinear for non-RGB images not to collapse into grayscale (#2089)

* * interpolate_bilinear uses now pixel_to_vector for shorter code.

* pixels now have operator!=.

* * Explicitely float interpolation

* Using C++11 static_assert() in interpolation.

* * Corrected documentation for interpolate_bilinear, interpolate_quadratic

* * Corrected formatting near interpolate_bilinear

Changed C-style cast to static_cast.

Reduce code duplication a bit and make equal_error_rate() give correct results when called on data where all detection scores are identical.

Previously it would say the EER was 0, but really it should have said 1 in this case.

* wip: attempt to use cuda for loss mse channel

* wip: maybe this is a step in the right direction

* Try to fix dereferencing the truth data (#1)

* Try to fix dereferencing the truth data

* Fix memory layout

* fix loss scaling and update tests

* rename temp1 to temp

* readd lambda captures for output_width and output_height

clangd was complaining about this, and suggested me to remove them
in the first, place:

```
Lambda capture 'output_height' is not required to be captured for this use (fix available)
Lambda capture 'output_width' is not required to be captured for this use (fix available)
```

* add a weighted_loss typedef to loss_multiclass_log_weighted_ for consistency

* update docs for weighted losses

* refactor multi channel loss and add cpu-cuda tests

* make operator() const

* make error relative to the loss value
Co-authored-by: Juha Reunanen <juha.reunanen@tomaattinen.com>