# LICENSE HEADER MANAGED BY add-license-header # # Copyright 2018 Kornia Team # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from __future__ import annotations from typing import Any, Optional import torch from kornia.augmentation import LongestMaxSize from kornia.augmentation.container.augment import AugmentationSequential from kornia.contrib.models import Prompts, SegmentationResults from kornia.contrib.models.sam import Sam, SamConfig from kornia.core import Tensor, pad, tensor from kornia.core.check import KORNIA_CHECK, KORNIA_CHECK_IS_TENSOR, KORNIA_CHECK_SHAPE from kornia.enhance import normalize from kornia.geometry.boxes import Boxes from kornia.geometry.keypoints import Keypoints class VisualPrompter: r"""Allow the user to run multiple query with multiple prompts for a model. At the moment, we just support the SAM model. The model is loaded based on the given config. For default the images are transformed to have their long side with size of the `image_encoder.img_size`. This Prompter class ensure to transform the images and the prompts before prediction. Also, the image is passed automatically for the method `preprocess_image`, which is responsible for normalize the image and pad it to have the right size for the SAM model :math:`(\text{image_encoder.img_size}, \text{image_encoder.img_size})`. For default the image is normalized by the mean and standard deviation of the SAM dataset values. Args: config: A model config to generate the model. Now just the SAM model is supported. device: The desired device to use the model. dtype: The desired dtype to use the model. Example: >>> # prompter = VisualPrompter() # Will load the vit h for default >>> # You can load a custom SAM type for modifying the config >>> prompter = VisualPrompter(SamConfig('vit_b')) >>> image = torch.rand(3, 25, 30) >>> prompter.set_image(image) >>> boxes = Boxes( ... torch.tensor( ... [[[[0, 0], [0, 10], [10, 0], [10, 10]]]], ... device=prompter.device, ... dtype=torch.float32 ... ), ... mode='xyxy' ... ) >>> prediction = prompter.predict(boxes=boxes) >>> prediction.logits.shape torch.Size([1, 3, 256, 256]) """ def __init__( self, config: Optional[SamConfig] = None, device: Optional[torch.device] = None, dtype: Optional[torch.dtype] = None, ) -> None: super().__init__() if config is None: config = SamConfig(model_type="vit_h", pretrained=True) if isinstance(config, SamConfig): self.model = Sam.from_config(config) transforms = (LongestMaxSize(self.model.image_encoder.img_size, p=1.0),) self.pixel_mean: Optional[Tensor] = tensor([123.675, 116.28, 103.53], device=device, dtype=dtype) / 255 self.pixel_std: Optional[Tensor] = tensor([58.395, 57.12, 57.375], device=device, dtype=dtype) / 255 else: raise NotImplementedError self.model = self.model.to(device=device, dtype=dtype) self.transforms = AugmentationSequential(*transforms, same_on_batch=True) self.device = device self.dtype = dtype self._original_image_size: None | tuple[int, int] = None self._input_image_size: None | tuple[int, int] = None self._input_encoder_size: None | tuple[int, int] = None self.reset_image() def preprocess_image(self, x: Tensor, mean: Optional[Tensor] = None, std: Optional[Tensor] = None) -> Tensor: """Normalize and pad a tensor. For normalize the tensor: will prioritize the `mean` and `std` passed as argument, if None will use the default Sam Dataset values. For pad the tensor: Will pad the tensor into the right and bottom to match with the size of `self.model.image_encoder.img_size` Args: x: The image to be preprocessed mean: Mean for each channel. std: Standard deviations for each channel. Returns: The image preprocessed (normalized if has mean and str available and padded to encoder size) """ if isinstance(mean, Tensor) and isinstance(std, Tensor): x = normalize(x, mean, std) elif isinstance(self.pixel_mean, Tensor) and isinstance(self.pixel_std, Tensor): x = normalize(x, self.pixel_mean, self.pixel_std) encoder_im_size = self.model.image_encoder.img_size pad_h = encoder_im_size - x.shape[-2] pad_w = encoder_im_size - x.shape[-1] x = pad(x, (0, pad_w, 0, pad_h)) return x @torch.no_grad() def set_image(self, image: Tensor, mean: Optional[Tensor] = None, std: Optional[Tensor] = None) -> None: """Set the embeddings from the given image with `image_decoder` of the model. Prepare the given image with the selected transforms and the preprocess method. Args: image: RGB image. Normally images with range of [0-1], the model preprocess normalize the pixel values with the mean and std defined in its initialization. Expected to be into a float32 dtype. Shape :math:`(3, H, W)`. mean: mean value of dataset for normalization. std: standard deviation of dataset for normalization. """ KORNIA_CHECK_SHAPE(image, ["3", "H", "W"]) self.reset_image() self._original_image_size = (image.shape[-2], image.shape[-1]) image = self.transforms(image, data_keys=["input"]) self._tfs_params = self.transforms._params self._input_image_size = (image.shape[-2], image.shape[-1]) image = self.preprocess_image(image, mean, std) self._input_encoder_size = (image.shape[-2], image.shape[-1]) self.image_embeddings = self.model.image_encoder(image) self.is_image_set = True def _valid_keypoints(self, keypoints: Keypoints | Tensor, labels: Tensor) -> Keypoints: """Validate the keypoints shape and ensure to be a Keypoints.""" KORNIA_CHECK_SHAPE(keypoints.data, ["K", "N", "2"]) KORNIA_CHECK_SHAPE(labels.data, ["K", "N"]) KORNIA_CHECK(keypoints.shape[0] == labels.shape[0], "The keypoints and labels should have the same batch size") if isinstance(keypoints, Tensor): keypoints = Keypoints.from_tensor(keypoints) return keypoints def _valid_boxes(self, boxes: Boxes | Tensor) -> Boxes: """Validate the boxes shape and ensure to be a Boxes into xyxy mode.""" if isinstance(boxes, Tensor): KORNIA_CHECK_SHAPE(boxes.data, ["K", "4"]) boxes = Boxes(boxes, mode="xyxy") if boxes.mode == "xyxy": boxes_xyxy = boxes else: boxes_xyxy = Boxes(boxes.to_tensor(mode="xyxy"), mode="xyxy") return boxes_xyxy def _valid_masks(self, masks: Tensor) -> Tensor: """Validate the input masks shape.""" KORNIA_CHECK_SHAPE(masks, ["K", "1", "256", "256"]) return masks def _transform_prompts( self, *prompts: Tensor | Boxes | Keypoints, data_keys: Optional[list[str]] = None ) -> dict[str, Tensor | Boxes | Keypoints]: transformed_prompts = self.transforms(*prompts, data_keys=data_keys, params=self._tfs_params) if data_keys is None: data_keys = [] # prevent unpacking tensor when creating the output dict (issue #2627) if not isinstance(transformed_prompts, (list, tuple)): transformed_prompts = [transformed_prompts] return {key: transformed_prompts[idx] for idx, key in enumerate(data_keys)} def preprocess_prompts( self, keypoints: Optional[Keypoints | Tensor] = None, keypoints_labels: Optional[Tensor] = None, boxes: Optional[Boxes | Tensor] = None, masks: Optional[Tensor] = None, ) -> Prompts: """Validate and preprocess the given prompts to be aligned with the input image.""" data_keys = [] to_transform: list[Keypoints | Boxes | Tensor] = [] if isinstance(keypoints, (Keypoints, Tensor)) and isinstance(keypoints_labels, Tensor): keypoints = self._valid_keypoints(keypoints, keypoints_labels) data_keys.append("keypoints") to_transform.append(keypoints) if isinstance(boxes, (Boxes, Tensor)): self._valid_boxes(boxes) data_keys.append("bbox_xyxy") to_transform.append(boxes) if isinstance(masks, Tensor): self._valid_masks(masks) data = self._transform_prompts(*to_transform, data_keys=data_keys) if "keypoints" in data and isinstance(data["keypoints"], Keypoints): kpts_tensor = data["keypoints"].to_tensor() if KORNIA_CHECK_IS_TENSOR(kpts_tensor) and KORNIA_CHECK_IS_TENSOR(keypoints_labels): points = (kpts_tensor, keypoints_labels) else: points = None if "bbox_xyxy" in data and isinstance(data["bbox_xyxy"], Boxes): _bbox = data["bbox_xyxy"].to_tensor(mode="xyxy") if KORNIA_CHECK_IS_TENSOR(_bbox): bbox = _bbox else: bbox = None return Prompts(points=points, boxes=bbox, masks=masks) @torch.no_grad() def predict( self, keypoints: Optional[Keypoints | Tensor] = None, keypoints_labels: Optional[Tensor] = None, boxes: Optional[Boxes | Tensor] = None, masks: Optional[Tensor] = None, multimask_output: bool = True, output_original_size: bool = True, ) -> SegmentationResults: """Predict masks for the given image based on the input prompts. Args: keypoints: Point prompts to the model. Each point is in (X,Y) in pixels. Shape :math:`(K, N, 2)`. Where `N` is the number of points and `K` the number of prompts. keypoints_labels: Labels for the point prompts. 1 indicates a foreground point and 0 indicates a background point. Shape :math:`(K, N)`. Where `N` is the number of points, and `K` the number of prompts. boxes: A box prompt to the model. If a tensor, should be in a xyxy mode. Shape :math:`(K, 4)` masks: A low resolution mask input to the model, typically coming from a previous prediction iteration. Has shape :math:`(K, 1, H, W)`, where for SAM, H=W=256. multimask_output: If true, the model will return three masks. For ambiguous input prompts (such as a single click), this will often produce better masks than a single prediction. If only a single mask is needed, the model's predicted quality score can be used to select the best mask. For non-ambiguous prompts, such as multiple input prompts, multimask_output=False can give better results. output_original_size: If true, the logits of `SegmentationResults` will be post-process to match the original input image size. Returns: A prediction with the logits and scores (IoU of each predicted mask) """ KORNIA_CHECK(self.is_image_set, "An image must be set with `self.set_image(...)` before `predict` be called!") prompts = self.preprocess_prompts(keypoints, keypoints_labels, boxes, masks) # Embed prompts sparse_embeddings, dense_embeddings = self.model.prompt_encoder( points=prompts.points, boxes=prompts.boxes, masks=prompts.masks ) del prompts # Predict masks logits, scores = self.model.mask_decoder( image_embeddings=self.image_embeddings, image_pe=self.model.prompt_encoder.get_dense_pe(), sparse_prompt_embeddings=sparse_embeddings, dense_prompt_embeddings=dense_embeddings, multimask_output=multimask_output, ) results = SegmentationResults(logits, scores) if ( output_original_size and isinstance(self._input_image_size, tuple) and isinstance(self._original_image_size, tuple) ): results.original_res_logits(self._input_image_size, self._original_image_size, self._input_encoder_size) # results = results.squeeze(0) return results def reset_image(self) -> None: self._tfs_params = None self._original_image_size = None self._input_image_size = None self._input_encoder_size = None if hasattr(self, "image_embeddings"): del self.image_embeddings self.image_embeddings = None self.is_image_set = False def compile( self, *, fullgraph: bool = False, dynamic: bool = False, backend: str = "inductor", mode: Optional[str] = None, options: Optional[dict[Any, Any]] = None, disable: bool = False, ) -> None: """Apply `torch.compile(...)`/dynamo API into the VisualPrompter API. .. note:: For more information about the dynamo API check the official docs https://pytorch.org/docs/stable/generated/torch.compile.html Args: fullgraph: Whether it is ok to break model into several subgraphs dynamic: Use dynamic shape tracing backend: backend to be used mode: Can be either “default”, “reduce-overhead” or “max-autotune” options: A dictionary of options to pass to the backend. disable: Turn torch.compile() into a no-op for testing Example: >>> # prompter = VisualPrompter() >>> # prompter.compile() # You should have torch >= 2.0.0 installed >>> # Use the prompter methods ... """ # self.set_image = torch.compile( # type: ignore[method-assign] # self.set_image, # fullgraph=fullgraph, # dynamic=dynamic, # backend=backend, # mode=mode, # options=options, # disable=disable, # ) # FIXME: compile set image will try to compile AugmentationSequential which fails self.model.image_encoder = torch.compile( # type: ignore self.model.image_encoder, fullgraph=fullgraph, dynamic=dynamic, backend=backend, mode=mode, options=options, disable=disable, ) # self.preprocess_image = torch.compile( # type: ignore[method-assign] # self.preprocess_image, # fullgraph=fullgraph, # dynamic=dynamic, # backend=backend, # mode=mode, # options=options, # disable=disable, # ) # FIXME: compile predict will try to compile Preproc prompts, which need to compileAugmentationSequential # which fails # self.predict = torch.compile( # type: ignore[method-assign] # self.predict, # fullgraph=fullgraph, # dynamic=dynamic, # backend=backend, # mode=mode, # options=options, # disable=disable, # ) self.model.mask_decoder = torch.compile( # type: ignore self.model.mask_decoder, fullgraph=fullgraph, dynamic=dynamic, backend=backend, mode=mode, options=options, disable=disable, ) self.model.prompt_encoder = torch.compile( # type: ignore self.model.prompt_encoder, fullgraph=fullgraph, dynamic=dynamic, backend=backend, mode=mode, options=options, disable=disable, )