o
V hR @ s d dl Z d dlZd dlmZmZ d dlmZ d dlmZ d dlm Z m
Z
m Z m Z m
Z
mZ d dlZd dlZd dlmZ ddlmZmZ d d
lmZ d d lmZmZ d d l m Z e
ejeje ej e ej f Z e
ejeje ej f Z d
Z!G dd dee Z"G dd de"Z#G dd de"Z$G dd de"Z%G dd de"Z&G dd de"Z'de(dej fd d Z)de(de
ej ej f fd d Z*dS ) N)ABCabstractmethod)globPath)AnyCallableListOptionalTupleUnion)Image )
decode_png read_file )default_loader) _read_pfmverify_str_arg)
VisionDataset) KittiFlowSintelFlyingThings3D FlyingChairsHD1Kc s e Zd ZdZdefdeeef dee de ege
f ddf fdd
Z d edee j e
jf fd
d Zed efd d
Zdedeeef fddZdefddZdede
jjjfddZ ZS ) FlowDatasetFNroot
transformsloaderreturnc s* t j|d || _g | _g | _|| _d S )N)r )super__init__r
_flow_list _image_list_loader)selfr r r __class__ V/var/www/vscode/kcb/lib/python3.10/site-packages/torchvision/datasets/_optical_flow.pyr! $ s
zFlowDataset.__init__ file_namec C s
| |S N)r$ r% r* r( r( r) _read_img2
zFlowDataset._read_imgc C s d S r+ r( r, r( r( r)
_read_flow5 s zFlowDataset._read_flowindexc C s | | j| d }| | j| d }| jr*| | j| }| jr'|\}}nd }nd }}| jd ur?| ||||\}}}}| jsF|d urL||||fS |||fS )Nr r )r- r# r" r/ _has_builtin_flow_maskr )r% r0 img1img2flowvalid_flow_maskr( r( r) __getitem__: s
zFlowDataset.__getitem__c C s
t | jS r+ )lenr# )r% r( r( r) __len__Q r. zFlowDataset.__len__vc C s t jj| g| S r+ )torchutilsdata
ConcatDataset)r% r9 r( r( r) __rmul__T s zFlowDataset.__rmul__) __name__
__module__ __qualname__r1 r r strr r
r r r! r
r: Tensorr- r r/ intT1T2r6 r8 r; r< r= r>
__classcell__r( r( r& r) r s&
r c
s e Zd ZdZdddefdeeef dededee d e ege
f d
df fd d
Z d
e d
ee
ef f fdd Zded
ejfddZ ZS )r a `Sintel `_ Dataset for optical flow.
The dataset is expected to have the following structure: ::
root
Sintel
testing
clean
scene_1
scene_2
...
final
scene_1
scene_2
...
training
clean
scene_1
scene_2
...
final
scene_1
scene_2
...
flow
scene_1
scene_2
...
Args:
root (str or ``pathlib.Path``): Root directory of the Sintel Dataset.
split (string, optional): The dataset split, either "train" (default) or "test"
pass_name (string, optional): The pass to use, either "clean" (default), "final", or "both". See link above for
details on the different passes.
transforms (callable, optional): A function/transform that takes in
``img1, img2, flow, valid_flow_mask`` and returns a transformed version.
``valid_flow_mask`` is expected for consistency with other datasets which
return a built-in valid mask, such as :class:`~torchvision.datasets.KittiFlow`.
loader (callable, optional): A function to load an image given its path.
By default, it uses PIL as its image loader, but users could also pass in
``torchvision.io.decode_image`` for decoding image data into tensors directly.
traincleanNr split pass_namer r r c
s t j|||d t|ddd t|ddd |dkr dd gn|g}t|d
}|d d }|D ]T}|d
kr8d n|}|| | } t| D ]>}
ttt| |
d } t t
| d D ]} | j | | | | d gg7 _ q[|d
kr| j ttt||
d 7 _ qEq0d S )Nr r r rJ rH test valid_valuesrK rI finalbothrS rI rR r trainingr4 rH *.pngr *.flo)
r r! r r oslistdirsortedr rB ranger7 r# r" )
r% r rJ rK r r passes flow_root split_dir
image_rootscene
image_listir& r( r) r! s$ ""zSintel.__init__r0 c t |S a Return example at given index.
Args:
index(int): The index of the example to retrieve
Returns:
tuple: A 3-tuple with ``(img1, img2, flow)``.
The flow is a numpy array of shape (2, H, W) and the images are PIL images.
``flow`` is None if ``split="test"``.
If a valid flow mask is generated within the ``transforms`` parameter,
a 4-tuple with ``(img1, img2, flow, valid_flow_mask)`` is returned.
r r6 r% r0 r& r( r) r6
zSintel.__getitem__r* c C t |S r+ _read_flor, r( r( r) r/ zSintel._read_flowr? r@ rA __doc__r r rB r r
r r r! rD rE rF r6 npndarrayr/ rG r( r( r& r) r X s* .
r c s e Zd ZdZdZddefdeeef dede e
de
ege f d df
fd
d
Z d e
d eeef f fd
d Zded eejejf fddZ ZS )r a `KITTI `__ dataset for optical flow (2015).
The dataset is expected to have the following structure: ::
root
KittiFlow
testing
image_2
training
image_2
flow_occ
Args:
root (str or ``pathlib.Path``): Root directory of the KittiFlow Dataset.
split (string, optional): The dataset split, either "train" (default) or "test"
transforms (callable, optional): A function/transform that takes in
``img1, img2, flow, valid_flow_mask`` and returns a transformed version.
loader (callable, optional): A function to load an image given its path.
By default, it uses PIL as its image loader, but users could also pass in
``torchvision.io.decode_image`` for decoding image data into tensors directly.
TrH Nr rJ r r r c s t j|||d t|ddd t|d |d }ttt|d d }ttt|d d }|r6|s:td
t||D ]\}}| j ||gg7 _ q?|d krattt|d d | _
d S d S )
NrL rJ rM rO r ingimage_2z*_10.pngz*_11.pngzZCould not find the Kitti flow images. Please make sure the directory structure is correct.rH flow_occ) r r! r r rY r rB FileNotFoundErrorzipr# r" ) r% r rJ r r images1images2r2 r3 r& r( r) r! s zKittiFlow.__init__r0 c rb )a Return example at given index.
Args:
index(int): The index of the example to retrieve
Returns:
tuple: A 4-tuple with ``(img1, img2, flow, valid_flow_mask)``
where ``valid_flow_mask`` is a numpy boolean mask of shape (H, W)
indicating which flow values are valid. The flow is a numpy array of
shape (2, H, W) and the images are PIL images. ``flow`` and ``valid_flow_mask`` are None if
``split="test"``.
rd re r& r( r) r6 rf zKittiFlow.__getitem__r* c C rg r+ )_read_16bits_png_with_flow_and_valid_maskr, r( r( r) r/ rj zKittiFlow._read_flow)r? r@ rA rl r1 r r rB r r
r r r! rD rE rF r6 r rm rn r/ rG r( r( r& r) r s&
&r c sr e Zd ZdZddeeef dedee ddf fdd
Z d
e
dee e f f fd d Z
d
edejfddZ ZS )r a `FlyingChairs `_ Dataset for optical flow.
You will also need to download the FlyingChairs_train_val.txt file from the dataset page.
The dataset is expected to have the following structure: ::
root
FlyingChairs
data
00001_flow.flo
00001_img1.ppm
00001_img2.ppm
...
FlyingChairs_train_val.txt
Args:
root (str or ``pathlib.Path``): Root directory of the FlyingChairs Dataset.
split (string, optional): The dataset split, either "train" (default) or "val"
transforms (callable, optional): A function/transform that takes in
``img1, img2, flow, valid_flow_mask`` and returns a transformed version.
``valid_flow_mask`` is expected for consistency with other datasets which
return a built-in valid mask, such as :class:`~torchvision.datasets.KittiFlow`.
rH Nr rJ r r c
s t j||d t|ddd t|d }ttt|d d }ttt|d d }d }tj || s;t
d
t j t|| t j
d }tt|D ]4}|| } |d kr[| d
ksc|dkr| dkr| j|| g7 _| j|d| |d| d
gg7 _qMd S )N)r r rJ )rH valrO r r< z*.ppmrV zFlyingChairs_train_val.txtzmThe FlyingChairs_train_val.txt file was not found - please download it from the dataset page (see docstring).)dtyperH r rx r )r r! r r rY r rB rW pathexistsrr rm loadtxtint32rZ r7 r" r# )
r% r rJ r imagesflowssplit_file_name
split_listra split_idr& r( r) r! s$ (zFlyingChairs.__init__r0 c rb )a Return example at given index.
Args:
index(int): The index of the example to retrieve
Returns:
tuple: A 3-tuple with ``(img1, img2, flow)``.
The flow is a numpy array of shape (2, H, W) and the images are PIL images.
``flow`` is None if ``split="val"``.
If a valid flow mask is generated within the ``transforms`` parameter,
a 4-tuple with ``(img1, img2, flow, valid_flow_mask)`` is returned.
rd re r& r( r) r6 * rf zFlyingChairs.__getitem__r* c C rg r+ rh r, r( r( r) r/ 9 rj zFlyingChairs._read_flow)rH N)r? r@ rA rl r rB r r
r r! rD rE rF r6 rm rn r/ rG r( r( r& r) r s
, r c s e Zd ZdZddddefdeeef deded ed
ee d e ege
f d df fd
d
Z de d ee
ef f fdd Zded ejfddZ ZS )r a `FlyingThings3D `_ dataset for optical flow.
The dataset is expected to have the following structure: ::
root
FlyingThings3D
frames_cleanpass
TEST
TRAIN
frames_finalpass
TEST
TRAIN
optical_flow
TEST
TRAIN
Args:
root (str or ``pathlib.Path``): Root directory of the intel FlyingThings3D Dataset.
split (string, optional): The dataset split, either "train" (default) or "test"
pass_name (string, optional): The pass to use, either "clean" (default) or "final" or "both". See link above for
details on the different passes.
camera (string, optional): Which camera to return images from. Can be either "left" (default) or "right" or "both".
transforms (callable, optional): A function/transform that takes in
``img1, img2, flow, valid_flow_mask`` and returns a transformed version.
``valid_flow_mask`` is expected for consistency with other datasets which
return a built-in valid mask, such as :class:`~torchvision.datasets.KittiFlow`.
loader (callable, optional): A function to load an image given its path.
By default, it uses PIL as its image loader, but users could also pass in
``torchvision.io.decode_image`` for decoding image data into tensors directly.
rH rI leftNr rJ rK camerar r r c s t j|||d t|ddd | }t|ddd dgdgddgd| }t d d
d d kr6d d
gn g}t|d }d} t||| D ]\} ttt || | d }
t fdd|
D }
ttt |d | d } t fdd| D } |
r| st
dt |
| D ]b\} }
ttt | d }ttt |
d }t t
|d D ]A}dkr| j|| ||d gg7 _| j|| g7 _qdkr| j||d || gg7 _| j||d g7 _qqqHd S )NrL rJ rM rO rK rQ frames_cleanpassframes_finalpassr )r rightrS rS r r r ) into_future into_pastz*/*c 3 s | ] }t | V qd S r+ r ).0 image_dir)r r( r) z s z*FlyingThings3D.__init__.. optical_flowc 3 s | ] }t | V qd S r+ r )r flow_dirr directionr( r) r } s zcCould not find the FlyingThings3D flow images. Please make sure the directory structure is correct.rU z*.pfmr r r )r r! r upperr itertoolsproductrY r rB rr rs rZ r7 r# r" )r% r rJ rK r r r r[ cameras
directions
image_dirs flow_dirsr r r~ r ra r& r r) r! ] sJ zFlyingThings3D.__init__r0 c rb rc rd re r& r( r) r6 rf zFlyingThings3D.__getitem__r* c C rg r+ )r r, r( r( r) r/ rj zFlyingThings3D._read_flowrk r( r( r& r) r = s0 "
3 r c s e Zd ZdZdZddefdeeef dede e
de
ege f d df
fd
d
Z d ed e
ejejf fd
dZded eeef f fdd Z ZS )r a `HD1K `__ dataset for optical flow.
The dataset is expected to have the following structure: ::
root
hd1k
hd1k_challenge
image_2
hd1k_flow_gt
flow_occ
hd1k_input
image_2
Args:
root (str or ``pathlib.Path``): Root directory of the HD1K Dataset.
split (string, optional): The dataset split, either "train" (default) or "test"
transforms (callable, optional): A function/transform that takes in
``img1, img2, flow, valid_flow_mask`` and returns a transformed version.
loader (callable, optional): A function to load an image given its path.
By default, it uses PIL as its image loader, but users could also pass in
``torchvision.io.decode_image`` for decoding image data into tensors directly.
TrH Nr rJ r r r c
sH t j|||d t|ddd t|d }|dkrktdD ]K}ttt|d d |d
d }ttt|d d
|d
d }tt|d D ] }| j || g7 _ | j
|| ||d gg7 _
qLq n0ttt|d d
d } ttt|d d
d }
t | |
D ]\} } | j
| | gg7 _
q| j
st dd S )NrL rJ rM rO hd1krH $ hd1k_flow_gtrq 06dz_*.png
hd1k_inputrp r hd1k_challengez*10.pngz*11.pngzTCould not find the HD1K images. Please make sure the directory structure is correct.)
r r! r r rZ rY r rB r7 r" r# rs rr )
r% r rJ r r seq_idxr r~ ra rt ru image1image2r& r( r) r! s* $$" z
HD1K.__init__r* c C rg r+ rv r, r( r( r) r/ rj zHD1K._read_flowr0 c rb )a Return example at given index.
Args:
index(int): The index of the example to retrieve
Returns:
tuple: A 4-tuple with ``(img1, img2, flow, valid_flow_mask)`` where ``valid_flow_mask``
is a numpy boolean mask of shape (H, W)
indicating which flow values are valid. The flow is a numpy array of
shape (2, H, W) and the images are PIL images. ``flow`` and ``valid_flow_mask`` are None if
``split="test"``.
rd re r& r( r) r6 rf zHD1K.__getitem__)r? r@ rA rl r1 r r rB r r
r r r! r rm rn r/ rD rE rF r6 rG r( r( r& r) r s&
&r r* r c C s t | dF}tj|ddd }|dkrtdttj|ddd}ttj|ddd}tj|d d
| | d}|||d
d
d dW d S 1 sNw Y d S )
z#Read .flo file in Middlebury formatrbc )counts PIEHz)Magic number incorrect. Invalid .flo filez s0 :[FDfL"