mvit_v2_s

torchvision.models.video.mvit_v2_s(*, weights: Optional[MViT_V2_S_Weights] = None, progress: bool = True, **kwargs: Any) → MViT

Constructs a small MViTV2 architecture from Multiscale Vision Transformers and MViTv2: Improved Multiscale Vision Transformers for Classification and Detection.

Warning

The video module is in Beta stage, and backward compatibility is not guaranteed.

Parameters:

• weights (MViT_V2_S_Weights, optional) – The pretrained weights to use. See MViT_V2_S_Weights below for more details, and possible values. By default, no pre-trained weights are used.

• progress (bool, optional) – If True, displays a progress bar of the download to stderr. Default is True.

• **kwargs – parameters passed to the torchvision.models.video.MViT base class. Please refer to the source code for more details about this class.

class torchvision.models.video.MViT_V2_S_Weights(value)

The model builder above accepts the following values as the weights parameter. MViT_V2_S_Weights.DEFAULT is equivalent to MViT_V2_S_Weights.KINETICS400_V1. You can also use strings, e.g. weights='DEFAULT' or weights='KINETICS400_V1'.

MViT_V2_S_Weights.KINETICS400_V1:

The weights were ported from the paper. The accuracies are estimated on video-level with parameters frame_rate=7.5, clips_per_video=5, and clip_len=16 Also available as MViT_V2_S_Weights.DEFAULT.

acc@1 (on Kinetics-400)	80.757
acc@5 (on Kinetics-400)	94.665
min_size	height=224, width=224
min_temporal_size	16
categories	abseiling, air drumming, answering questions, … (397 omitted)
recipe	link
num_params	34537744
GFLOPS	64.22
File size	131.9 MB

The inference transforms are available at MViT_V2_S_Weights.KINETICS400_V1.transforms and perform the following preprocessing operations: Accepts batched (B, T, C, H, W) and single (T, C, H, W) video frame torch.Tensor objects. The frames are resized to resize_size=[256] using interpolation=InterpolationMode.BILINEAR, followed by a central crop of crop_size=[224, 224]. Finally the values are first rescaled to [0.0, 1.0] and then normalized using mean=[0.45, 0.45, 0.45] and std=[0.225, 0.225, 0.225]. Finally the output dimensions are permuted to (..., C, T, H, W) tensors.