#790 Add Ops\Boxes #867
#790 Add Ops\Boxes #867
Conversation
| return NULL; | ||
| } | ||
| #else | ||
| lib = dlopen((libName + ".so").c_str(), RTLD_LAZY); |
There was a problem hiding this comment.
This won't work on Mac, I believe. There, it's '.dylib'
| /// <param name="input">The first input tensor</param> | ||
| /// <param name="other">The second input tensor</param> | ||
| public static Tensor maximum(Tensor input, Tensor other) => input.maximum(other); | ||
| public static Tensor max(Tensor input, Tensor other) => maximum(input, other); |
There was a problem hiding this comment.
I don't think 'max' and 'maximum' are aliases of each other.
https://pytorch.org/docs/stable/generated/torch.max.html?highlight=max#torch.max
https://pytorch.org/docs/stable/generated/torch.maximum.html?highlight=maximum#torch.maximum
There was a problem hiding this comment.
there is an overload at the bottom of the 'max‘ page
torch.max(input, other, *, out=None) → Tensor
See torch.maximum().
| /// <param name="input">The first input tensor</param> | ||
| /// <param name="other">The second input tensor</param> | ||
| public static Tensor minimum(Tensor input, Tensor other) => input.minimum(other); | ||
| public static Tensor min(Tensor input, Tensor other) => minimum(input, other); |
There was a problem hiding this comment.
See comment on 'max'
| /// </summary> | ||
| public static class DeconstructExtension | ||
| { | ||
| public static void Deconstruct<T>(this IEnumerable<T> seq, out T first, out IEnumerable<T> rest) |
There was a problem hiding this comment.
Doc comments on all public APIs, please.
| var scores = torch.from_array(new[] { 0.9 }); | ||
|
|
||
| var nms_boxes = nms(boxes, scores); | ||
| var nms_boxes = nms_custom(boxes, scores); |
There was a problem hiding this comment.
This means there's no unit test for 'nms' -- please add one.
|
@huangyangyu -- can you take a look at this PR? You added the C# implementation of 'nms' |
|
I'm curious about libtorchvision.dll -- you uploaded it to Google drive, but we'll have to come up with a more automated way of building it. |
|
I'm wondering if it wouldn't be more convenient, in the long run to just add torchvision (and audio, and text) as linked projects and build the native code into our native project. |
|
@NiklasGustafsson it will be great if it's possible to build torchvision automatically, but currently I cannot find a way to do it. |
|
Currently my branch is a mess, I'll consider create pull request later. |
2 participants
Add nms method from torch vision.
Build torchvision dll from source and rename to "libtorchvision", put it and all dependencies to the same path of exe file.
Tested on windows, the dll uploaded to https://drive.google.com/drive/folders/14R6_HzDEOlespcOd6pAWYgxjoYjD1Xpz?usp=share_link
command to build torch vision: