^jaGdZddlmZddlmZmZmZddlZddl m Z ddl m Z ddl mZddlmZmZmZmZmZddlmZdd lmZmZmZdd lmZdd lmZm Z dd l!m"Z"m#Z#d gZ$Gdde Z%Gdd e Z&y)aTransform classes for dropout-based augmentations. This module contains transform classes for various dropout techniques used in image augmentation. It provides the base dropout class and specialized implementations like PixelDropout. These transforms randomly remove or modify pixels, channels, or regions in images, which can help models become more robust to occlusions and missing information. ) annotations)AnyLiteralcastN)get_num_channels)Field) functional)cutoutcutout_on_volumecutout_on_volumesfilter_bboxes_by_holesfilter_keypoints_in_holes) BboxProcessordenormalize_bboxesnormalize_bboxes)KeypointsProcessor)BaseTransformInitSchema DualTransform) ALL_TARGETSTargets PixelDropoutceZdZUdZeZded<GddeZ dfd Z ddZ ddZ dd Z dd Z dd Zdd Zdd Z ddZ ddZddZxZS) BaseDropoutaTBase class for dropout-style transformations. This class provides common functionality for various dropout techniques, including applying cutouts to images and masks. Args: fill (tuple[float, ...] | float | Literal["random", "random_uniform", "inpaint_telea", "inpaint_ns"]): Value to fill dropped regions. fill_mask (tuple[float, ...] | float | None): Value to fill dropped regions in the mask. If None, the mask is not modified. p (float): Probability of applying the transform. Targets: image, mask, bboxes, keypoints, volume, mask3d Image types: uint8, float32 Examples: >>> import numpy as np >>> import albumentations as A >>> >>> # Example of a custom dropout transform inheriting from BaseDropout >>> class CustomDropout(A.BaseDropout): ... def __init__(self, num_holes_range=(4, 8), hole_size_range=(10, 20), *args, **kwargs): ... super().__init__(*args, **kwargs) ... self.num_holes_range = num_holes_range ... self.hole_size_range = hole_size_range ... ... def get_params_dependent_on_data(self, params, data): ... img = data["image"] ... height, width = img.shape[:2] ... ... # Generate random holes ... num_holes = self.py_random.randint(*self.num_holes_range) ... hole_sizes = self.py_random.randint(*self.hole_size_range, size=num_holes) ... ... holes = [] ... for i in range(num_holes): ... # Random position for each hole ... x1 = self.py_random.randint(0, max(1, width - hole_sizes[i])) ... y1 = self.py_random.randint(0, max(1, height - hole_sizes[i])) ... x2 = min(width, x1 + hole_sizes[i]) ... y2 = min(height, y1 + hole_sizes[i]) ... holes.append([x1, y1, x2, y2]) ... ... # Return holes and random seed ... return { ... "holes": np.array(holes) if holes else np.empty((0, 4), dtype=np.int32), ... "seed": self.py_random.integers(0, 100000) ... } >>> >>> # Prepare sample data >>> image = np.random.randint(0, 256, (100, 100, 3), dtype=np.uint8) >>> mask = np.random.randint(0, 2, (100, 100), dtype=np.uint8) >>> bboxes = np.array([[0.1, 0.1, 0.4, 0.4], [0.6, 0.6, 0.9, 0.9]]) >>> >>> # Create a transform with custom dropout >>> transform = A.Compose([ ... CustomDropout( ... num_holes_range=(3, 6), # Generate 3-6 random holes ... hole_size_range=(5, 15), # Holes of size 5-15 pixels ... fill=0, # Fill holes with black ... fill_mask=1, # Fill mask holes with 1 ... p=1.0 # Always apply for this example ... ) ... ], bbox_params=A.BboxParams(format='yolo', min_visibility=0.3)) >>> >>> # Apply the transform >>> transformed = transform(image=image, mask=mask, bboxes=bboxes) >>> >>> # Get the transformed data >>> dropout_image = transformed["image"] # Image with random holes filled with 0 >>> dropout_mask = transformed["mask"] # Mask with same holes filled with 1 >>> dropout_bboxes = transformed["bboxes"] # Bboxes filtered by visibility threshold ztuple[Targets, ...] | Targets_targetsc"eZdZUded<ded<y)BaseDropout.InitSchema^tuple[float, ...] | float | Literal['random', 'random_uniform', 'inpaint_telea', 'inpaint_ns']fill tuple[float, ...] | float | None fill_maskN)__name__ __module__ __qualname____annotations__z/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/albumentations/augmentations/dropout/transforms.py InitSchemarssll33r&r(cBt||||_||_yN)p)super__init__rr )selfrr r+ __class__s r'r-zBaseDropout.__init__ws# 1 "r&c |jdk(r|S|jdvrt|}|dvr tdt |||jt j j|S)Nr> inpaint_ns inpaint_telea/Inpainting works only for 1 or 3 channel images)sizerr ValueErrorr nprandom default_rng)r.imgholesseedparams num_channelss r'applyzBaseDropout.applysa ::?J 997 7+C0L6) !RSSc5$))RYY-B-B4-HIIr&c |jdk(r|S|jdvr/|jdk(r|jdnd}|dvr t dt |||jt jj|S)Nr>r1r2r5r4r3r6) r7rndimshaper8r r9r:r;)r.imagesr=r>r?r@s r'apply_to_imageszBaseDropout.apply_to_imagessr ::?M 997 7.4kkQ.>6<<?AL6) !RSStyy")):O:OPT:UVVr&c ,|j|||fi|S)N)rG)r.volumer=r>r?s r'apply_to_volumezBaseDropout.apply_to_volumes $t##FE4B6BBr&c |jdk(r|S|jdvr/|jdk(r|jdnd}|dvr t dt |||jt jj|S)Nr>r1r2rCr4r3r6) r7rrDrEr8r r9r:r;)r.volumesr=r>r?r@s r'apply_to_volumeszBaseDropout.apply_to_volumesss ::?N 997 7/6||q/@7==+aL6) !RSS %BIIr?s r'apply_to_mask3dzBaseDropout.apply_to_mask3ds@ >> !UZZ1_KeT^^RYY=R=RSW=XYYr&c |j|jdk(r|St|||jtjj |SrP)r r7r r9r:r;rQs r'apply_to_masks3dzBaseDropout.apply_to_masks3ds@ >> !UZZ1_K udnnbii>S>STX>YZZr&c |j|jdk(r|St|||jtjj |SrP)r r7r r9r:r;rQs r' apply_to_maskzBaseDropout.apply_to_masks? >> !UZZ1_KdE4>>2993H3H3NOOr&c |jdk(r|Std|jd}||S|ddd}t||}t t ||||j j|j j|S)NrrbboxesrE)min_areamin_visibility) r7r get_processorrrr r?r[r\)r.rYr=r? processor image_shapedenormalized_bboxess r'apply_to_bboxeszBaseDropout.apply_to_bboxess ::?M$*<*>     r&c |jdk(r|Std|jd}||jjs|St ||S)Nrr keypoints)r7rr]r?remove_invisibler)r.rcr=r?r^s r'apply_to_keypointszBaseDropout.apply_to_keypointssT ::? -t/A/A+/NO  I$4$4$E$E (E::r&ctd)Nz&Subclasses must implement this method.)NotImplementedError)r.r?datas r'get_params_dependent_on_dataz(BaseDropout.get_params_dependent_on_datas!"JKKr&)rrr rr+float) r< np.ndarrayr=rkr>intr?rreturnrk) rFrkr=rkr>rlr?rrmrk) rIrkr=rkr>rlr?rrmrk) rMrkr=rkr>rlr?rrmrk) rRrkr=rkr>rlr?rrmrk)rYrkr=rkr?rrmrk)rcrkr=rkr?rrmrkr?dict[str, Any]rhrormro)r!r"r#__doc__rrr$rr(r-rArGrJrNrSrUrWrareri __classcell__r/s@r'rr"sL\/:H+94,4#l#4#  #JWC YZ [ P        4 ; ; ; ;  ;Lr&rceZdZdZGddeZeZ d d fd Z d dZ d dZ ddZ ddZ dd Z xZS)rap Drops random pixels from the image. This transform randomly sets pixels in the image to a specified value, effectively "dropping out" those pixels. It can be applied to both the image and its corresponding mask. Args: dropout_prob (float): Probability of dropping out each pixel. Should be in the range [0, 1]. Default: 0.01 per_channel (bool): If True, the dropout mask will be generated independently for each channel. If False, the same dropout mask will be applied to all channels. Default: False drop_value (float | tuple[float, ...] | None): Value to assign to the dropped pixels. If None, the value will be randomly sampled for each application: - For uint8 images: Random integer in [0, 255] - For float32 images: Random float in [0, 1] If a single number, that value will be used for all dropped pixels. If a sequence, it should contain one value per channel. Default: 0 mask_drop_value (float | tuple[float, ...] | None): Value to assign to dropped pixels in the mask. If None, the mask will remain unchanged. If a single number, that value will be used for all dropped pixels in the mask. If a sequence, it should contain one value per channel. Default: None p (float): Probability of applying the transform. Should be in the range [0, 1]. Default: 0.5 Targets: image, mask, bboxes, keypoints, volume, mask3d Image types: uint8, float32 Note: - When applied to bounding boxes, this transform may cause some boxes to have zero area if all pixels within the box are dropped. Such boxes will be removed. - When applied to keypoints, keypoints that fall on dropped pixels will be removed if the keypoint processor is configured to remove invisible keypoints. Examples: >>> import numpy as np >>> import albumentations as A >>> image = np.random.randint(0, 256, (100, 100, 3), dtype=np.uint8) >>> mask = np.random.randint(0, 2, (100, 100), dtype=np.uint8) >>> transform = A.PixelDropout(dropout_prob=0.1, per_channel=True, p=1.0) >>> result = transform(image=image, mask=mask) >>> dropped_image, dropped_mask = result['image'], result['mask'] cJeZdZUeddZded<ded<ded <ded <y ) PixelDropout.InitSchemarr4)gelerj dropout_probbool per_channelr drop_valuemask_drop_valueN)r!r"r#rrxr$r%r&r'r(rus%#qQ/ e/4499r&r(c^t||||_||_||_||_yr*)r,r-rxrzr{r|)r.rxrzr{r|r+r/s r'r-zPixelDropout.__init__s4 1(&$.r&c 0tj|||S)aApply pixel dropout to the image. Args: img (np.ndarray): The image to apply the transform to. drop_mask (np.ndarray): The dropout mask. drop_values (np.ndarray): The values to assign to the dropped pixels. **params (Any): Additional parameters for the transform. Returns: np.ndarray: The transformed image. )fpixel pixel_dropout)r.r< drop_mask drop_valuesr?s r'rAzPixelDropout.apply*s&##CK@@r&c L|j|Stj|||S)aApply pixel dropout to the mask. Args: mask (np.ndarray): The mask to apply the transform to. mask_drop_mask (np.ndarray): The dropout mask for the mask. mask_drop_values (float | np.ndarray): The values to assign to the dropped pixels in the mask. **params (Any): Additional parameters for the transform. Returns: np.ndarray: The transformed mask. )r|rr)r.rRmask_drop_maskmask_drop_valuesr?s r'rWzPixelDropout.apply_to_mask?s+&    'K##D.:JKKr&c | |jr|Std|jd}||S|ddd}t||}|jrjt |j dkDrRt j||j ddkrdnd }|jd k(r|j dd k(r|d}n|}tj||||jj|jj}t||S) auApply pixel dropout to the bounding boxes. Args: bboxes (np.ndarray): The bounding boxes to apply the transform to. drop_mask (np.ndarray | None): The dropout mask for the bounding boxes. **params (Any): Additional parameters for the transform. Returns: np.ndarray: The transformed bounding boxes. NrrYrErZrCr)axisr5r4)rzrr]rlenrEr9anyrDfdropoutmask_dropout_bboxesr?r[r\r) r.rYrr?r^r_r` combined_maskresults r'razPixelDropout.apply_to_bboxesWs "   0 0M$*<*+>q+AQ+F -a 0 %M--       % %    + +    44r&c |S)aApply pixel dropout to the keypoints. Args: keypoints (np.ndarray): The keypoints to apply the transform to. **params (Any): Additional parameters for the transform. Returns: np.ndarray: The transformed keypoints. r%)r.rcr?s r'rezPixelDropout.apply_to_keypointss r&c@d|vr|dn|dd}tj|j|j|j|j }tj ||j|j }d}d}tj|}|jm|ktj|j|j|j|j }tj ||j|j }||||nd||dSddS)a(Generate parameters for pixel dropout based on input data. Args: params (dict[str, Any]): Transform parameters data (dict[str, Any]): Input data dictionary Returns: dict[str, Any]: Dictionary of parameters for applying the transform imagerFrN)rrrr) r get_drop_maskrErzrxrandom_generatorprepare_drop_valuesr{get_mask_arrayr|) r.r?rhreference_arrayrrrrrRs r'riz)PixelDropout.get_params_dependent_on_datas8,3d?$w-Xq@Q((  ! !        ! !  00  OO  ! ! $$T*    +0@#11   !!%% N &99$$%%  #&0>0JnPT4D4P 0   W[   r&)g{Gz?FrNg?) rxrjrzryr{rr|rr+rj) r<rkrrkrrkr?rrmrk) rRrkrrkrzfloat | np.ndarrayr?rrmrk)rYrkrznp.ndarray | Noner?rrmrk)rcrkr?rrmrkrn)r!r"r#rprr(rrr-rArWrarerirqrrs@r'rrsG3j:,: H#!78<@ / / /5 / : /  /A AA A  A  A*LL#L- L  L  L0/5/5%/5 /5  /5b  "4 4 4   4 r&)'rp __future__rtypingrrrnumpyr9albucorerpydanticr$albumentations.augmentations.dropoutr r/albumentations.augmentations.dropout.functionalr r r r r"albumentations.augmentations.pixelralbumentations.core.bbox_utilsrrr#albumentations.core.keypoints_utilsr(albumentations.core.transforms_interfacerr$albumentations.core.type_definitionsrr__all__rrr%r&r'rsg#%%%GD^^B[E  yL-yLxo =o r&