%`ibddlmZddlmZmZmZddlZddlmZddl m Z Gdde Z dS) ) annotations)AnyMappingSequenceN) tree_flatten)TVTensorczeZdZUdZded<eddd"d Zd d d d d#dZe d$d%dZd dd&d!Z d S)' KeyPointsu! :class:`torch.Tensor` subclass for tensors with shape ``[..., 2]`` that represent points in an image. .. note:: Support for keypoints was released in TorchVision 0.23 and is currently a BETA feature. We don't expect the API to change, but there may be some rare edge-cases. If you find any issues, please report them on our bug tracker: https://github.com/pytorch/vision/issues?q=is:open+is:issue Each point is represented by its X and Y coordinates along the width and height dimensions, respectively. Each point is represented by its X and Y coordinates along the width and height dimensions, respectively. KeyPoints may represent any object that can be represented by sequences of 2D points: - `Polygonal chains `_, including polylines, Bézier curves, etc., which can be of shape ``[N_chains, N_points, 2]``. - Polygons, which can be of shape ``[N_polygons, N_points, 2]``. - Skeletons, which can be of shape ``[N_skeletons, N_bones, 2, 2]`` for pose-estimation models. .. note:: Like for :class:`torchvision.tv_tensors.BoundingBoxes`, there should only be a single instance of the :class:`torchvision.tv_tensors.KeyPoints` class per sample e.g. ``{"img": img, "poins_of_interest": KeyPoints(...)}``, although one :class:`torchvision.tv_tensors.KeyPoints` object can contain multiple key points Args: data: Any data that can be turned into a tensor with :func:`torch.as_tensor`. canvas_size (two-tuple of ints): Height and width of the corresponding image or video. dtype (torch.dtype, optional): Desired data type of the bounding box. If omitted, will be inferred from ``data``. device (torch.device, optional): Desired device of the bounding box. If omitted and ``data`` is a :class:`torch.Tensor`, the device is taken from it. Otherwise, the bounding box is constructed on the CPU. requires_grad (bool, optional): Whether autograd should record operations on the bounding box. If omitted and ``data`` is a :class:`torch.Tensor`, the value is taken from it. Otherwise, defaults to ``False``. tuple[int, int] canvas_sizeT) check_dimstensor torch.Tensorrboolreturnc|rI|jdkr|d}n(|jddkrtd|j||}||_|S)Nrrz)Expected a tensor of shape (..., 2), not )ndim unsqueezeshape ValueError as_subclassr )clsrr rpointss u/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/torchvision/tv_tensors/_keypoints.py_wrapzKeyPoints._wrap;sx  ]{a))!,,b!Q&& ![V\![![\\\##C((( Ndtypedevice requires_graddatarr!torch.dtype | Noner"torch.device | str | int | Noner# bool | Nonecb|||||}|||S)Nr r ) _to_tensorr)rr$r r!r"r#rs r__new__zKeyPoints.__new__Fs4E&P]^^yy[y999routputargs Sequence[Any]kwargsMapping[str, Any] | Nonect||r!t|ndz\}}td|D}|jt |t jr3t |tst |d}n@t |ttfr$t|fd|D}|S)Nr,c3DK|]}t|t|VdS)N) isinstancer ).0xs r z)KeyPoints._wrap_output..[s2(\(\q:aQZC[C[(\(\(\(\(\(\(\rFr rc3RK|]!}t|dV"dS)Fr8N)r r)r5partr s rr7z)KeyPoints._wrap_output..bs7!v!vgk)//$K\a/"b"b!v!v!v!v!v!vr) rtuplevaluesnextr r4torchTensorr rlisttype)rr-r.r0 flat_params_first_keypoints_from_argsr s @r _wrap_outputzKeyPoints._wrap_outputRs&d.VeFMMOO.D.D.DTV&WXX Q$((\(\K(\(\(\$\$\!/; fel + + wJvy4Q4Q w__VQV_WWFF  . . w!T&\\!v!v!v!vou!v!v!vvvF r)tensor_contentsrFstrc8||jS)Nr)) _make_reprr )selfrFs r__repr__zKeyPoints.__repr__es4+;<<torch.utils._pytreer _tv_tensorr r r,rrrVs"""""")))))))))) ,,,,,, [=[=[=[=[=[=[=[=[=[=r