Haiping
The `iou_mask` is `NDArray`.  It means we use different ways in different dimension to select elements. One approach my idea is: Define a `IIndex` interface, `NDArray` and `Slice` implement...
Not only for masking, but also for any 1d array. We don’t have to have int to IIndex implicitly, we can use np.array(1) as alternative. The goal is having people...
@henon This situation doesn't work: `iou_mask` is `NDArray`, `yind` and `xind` are `int`:  
Remove NumSharp reference, just reference TensorFlow.NET project. It will include NumSharp automatically.
Remove project reference means remove it from package.  You still need: ```csharp using Numsharp; ``` The easiest step is just follow this [sample project](https://github.com/SciSharp/SciSharp-Stack-Examples).
@Nucs Do you have any plan on `String` support?
Should use Pandas.NET
@aidevnn Join us, build Deep Learning library based on current work.
@aidevnn Is `NDProxy` like a kind of Expression/ Func/ Delegate ? It can be executed after all operations have been decided ?
https://docs.scipy.org/doc/numpy/reference/generated/numpy.rot90.html