[jR X d dl Z d dlZd dlmZmZ d dlmZ d dlmZ d dlm Z m
Z
m Z m Z d dl
Zd dlZd dlmZ ddlmZmZ d d
lmZ d d lmZmZ d d lmZ e ej" ej" e ej: e ej: f Z e ej" ej" e ej: f Z d
Z G d dee Z! G d de! Z" G d de! Z# G d de! Z$ G d de! Z% G d de! Z&de'dej: fd Z(de'de ej: ej: f fd Z)y) N)ABCabstractmethod)globPath)AnyCallableOptionalUnion)Image )
decode_png read_file )default_loader) _read_pfmverify_str_arg)
VisionDataset) KittiFlowSintelFlyingThings3D FlyingChairsHD1Kc e Zd ZdZdefdeeef dee de ege
f ddf fd
Z dedee j e
j f fd Zedefd
Zd edeeef fd Zdefd
Zdede
j. j0 j2 fdZ xZS ) FlowDatasetFNroot
transformsloaderreturnc ^ t | | || _ g | _ g | _ || _ y )N)r )super__init__r
_flow_list _image_list_loader)selfr r r __class__s a/var/www/html/ia-enhanced/venv/lib/python3.12/site-packages/torchvision/datasets/_optical_flow.pyr" zFlowDataset.__init__$ s2 d#$%',. file_namec $ | j | S N)r% r& r* s r( _read_imgzFlowDataset._read_img2 s ||I&&r) c y r, r- s r(
_read_flowzFlowDataset._read_flow5 s
r) indexc | j | j | d }| j | j | d }| j r3| j | j | }| j r|\ }}nd }nd x}}| j
| j |||| \ }}}}| j s|||||fS |||fS )Nr r )r. r$ r# r1 _has_builtin_flow_maskr )r& r2 img1img2flowvalid_flow_masks r( __getitem__zFlowDataset.__getitem__: s ~~d . .u 5a 89~~d . .u 5a 89 ????4??5#9:D**(,%o"&%) )D? ?? &04dDRa0b -D$o & &/*Et_4 4t# #r) c , t | j S r, )lenr$ )r& s r( __len__zFlowDataset.__len__Q s 4##$$r) vc \ t j j j | g|z S r, )torchutilsdata
ConcatDataset)r& r= s r( __rmul__zFlowDataset.__rmul__T s# {{ --tfqj99r) ) __name__
__module__ __qualname__r4 r r strr r
r r r" r r? Tensorr. r r1 intT1T2r9 r< r@ rA rB rC
__classcell__r' s @r( r r s #
*.'5 CI X& #$
'3 '5ell1J+K '
C
$ $r2v $.% %:# :%++"2"2"@"@ :r) r c
e Zd ZdZdddefdeeef dededee d e ege
f d
df fd
Z d e d
ee
ef f fd
Zded
ej" fdZ xZS )r a `Sintel `_ Dataset for optical flow.
The dataset is expected to have the following structure: ::
root
Sintel
testing
clean
scene_1
scene_2
...
final
scene_1
scene_2
...
training
clean
scene_1
scene_2
...
final
scene_1
scene_2
...
flow
scene_1
scene_2
...
Args:
root (str or ``pathlib.Path``): Root directory of the Sintel Dataset.
split (string, optional): The dataset split, either "train" (default) or "test"
pass_name (string, optional): The pass to use, either "clean" (default), "final", or "both". See link above for
details on the different passes.
transforms (callable, optional): A function/transform that takes in
``img1, img2, flow, valid_flow_mask`` and returns a transformed version.
``valid_flow_mask`` is expected for consistency with other datasets which
return a built-in valid mask, such as :class:`~torchvision.datasets.KittiFlow`.
loader (callable, optional): A function to load an image given its path.
By default, it uses PIL as its image loader, but users could also pass in
``torchvision.io.decode_image`` for decoding image data into tensors directly.
traincleanNr split pass_namer r r c B t
| ||| t |dd t |dd |dk( rdd gn|g}t | d
z }|d z d z }|D ] }|d
k( rd n|}||z |z } t j
| D ] }
t
t t | |
z dz } t t | dz
D ]# } | xj | | | | dz ggz
c_ % |d
k( si| xj t
t t ||
z dz z
c_ y )Nr r r rQ rO test valid_valuesrR rP finalbothr[ rP rZ r trainingr7 rO *.pngr *.flo)
r! r" r r oslistdirsortedr rG ranger; r$ r# )r& r rQ rR r r passes flow_root split_dir
image_rootscene
image_listir' s r( r" zSintel.__init__ sA dz&Iug4EFy+`__ 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.
TrO Nr rQ r r r c t | ||| t |dd t | dz |dz z }t t t
|dz dz }t t t
|dz d z }|r|s t d
t || D ] \ }}| xj ||ggz
c_ |d k( r)t t t
|d z dz | _
y y )
NrT rQ rU rW r ingimage_2z*_10.pngz*_11.pngzZCould not find the Kitti flow images. Please make sure the directory structure is correct.rO flow_occ) r! r" r r ra r rG FileNotFoundErrorzipr$ r# )
r& r rQ r r images1images2r5 r6 r' s
r( r" zKittiFlow.__init__ s dz&Iug4EFDzK'55=9c$"2Z"? @ABc$"2Z"? @ABg#l
gw/ /JD$ $ . / G $T#dZ.?*.L*M%N ODO r) r2 c " t | | S )a Return example at given index.
Args:
index(int): The index of the example to retrieve
Returns:
tuple: A 4-tuple with ``(img1, img2, flow, valid_flow_mask)``
where ``valid_flow_mask`` is a numpy boolean mask of shape (H, W)
indicating which flow values are valid. The flow is a numpy array of
shape (2, H, W) and the images are PIL images. ``flow`` and ``valid_flow_mask`` are None if
``split="test"``.
rl rm s r( r9 zKittiFlow.__getitem__ rn r) r* c t | S r, )_read_16bits_png_with_flow_and_valid_maskr- s r( r1 zKittiFlow._read_flow
8CCr) )rD rE rF rt r4 r r rG r r
r r r" rI rJ rK r9 tupleru rv r1 rL rM s @r( r r s , "
)-'5 PCI P P X& P
#$ P
P4
*
*r2v
* DC DE"**bjj2H,I Dr) r c e Zd ZdZd deeef dedee ddf fd
Z de
dee e f f fd Z
d
edej fd Z xZS )
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`.
Nr rQ r r c t
| || t |dd t | dz }t t t
|dz dz }t t t
|dz dz }d }t j j ||z s t d
t j t
||z t j }t t | D ]] }|| } |d k( r| d
k( s |dk( s| dk( s | xj || gz
c_ | xj" |d|z |d|z d
z ggz
c_ _ y )N)r r rQ )rO valrW r rA z*.ppmr^ zFlyingChairs_train_val.txtzmThe FlyingChairs_train_val.txt file was not found - please download it from the dataset page (see docstring).)dtyperO r r r
)r! r" r r ra r rG r_ pathexistsr| ru loadtxtint32rb r; r# r$ ) r& r rQ r imagesflowssplit_file_name
split_listri split_idr' s r( r" zFlyingChairs.__init__ s4
dz:ug4DEDzN*S !8 9:;tCv
7 89:6ww~~d_ 45#
ZZD?$: ;288L
s5z" IA!!}H X]8WX=E!H:- fQUmVAEAI5F%G$HH Ir) r2 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="val"``.
If a valid flow mask is generated within the ``transforms`` parameter,
a 4-tuple with ``(img1, img2, flow, valid_flow_mask)`` is returned.
rl rm s r( r9 zFlyingChairs.__getitem__* rn r) r* c t | S r, rp r- s r( r1 zFlyingChairs._read_flow9 rr r) )rO N)rD rE rF rt r rG r r
r r" rI rJ rK r9 ru rv r1 rL rM s @r( r r sq 2IU39 - Ic IQYZbQc Ios I.
*
*r2v
* $C $BJJ $r) r c e Zd ZdZddddefdeeef deded ed
ee d e ege
f d df fd
Z de d ee
ef f fd Zded ej" fdZ xZS )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.
rO rP leftNr rQ rR camerar r r c
& t | ||| t |dd |j }t |dd dgdgddgd| }t d d
d k( rd d
gng}t | dz }d} t j ||| D ]~ \ }t t t ||z |z dz }
t fd|
D }
t t t |dz |z dz } t fd| D } |
r| s t d t |
| D ] \ } }
t t t | dz }t t t |
dz }t t | dz
D ] }dk( r;| xj || ||dz ggz
c_ | xj || gz
c_ Cdk( sI| xj ||dz || ggz
c_ | xj ||dz gz
c_ y )NrT rQ rU rW rR rY frames_cleanpassframes_finalpassr )r rightr[ r[ r r r ) into_future into_pastz*/*c 3 : K | ] }t | z y wr, r ).0 image_dirr s r( z*FlyingThings3D.__init__..z s UYY& 8 Us optical_flowc 3 @ K | ] }t | z z y wr, r )r flow_dirr directions r( r z*FlyingThings3D.__init__..} s ]xtH~ 9F B ]s zcCould not find the FlyingThings3D flow images. Please make sure the directory structure is correct.r] z*.pfmr r r )r! r" r upperr itertoolsproductra r rG r| r} rb r; r$ r# )r& r rQ rR r r r rc cameras
directions
image_dirs flow_dirsr r r r ri r r' s ` @r( r" zFlyingThings3D.__init__] s1 dz&Iug4EF
y+? tC7(:$; <= s5zA~. :A M1 ((fQiA -G,HH( E!H:5"k1 ((fQUmVAY-G,HH( E!a%L>9
:
: :r) r2 c " t | | S rk rl rm s r( r9 zFlyingThings3D.__getitem__ rn r) r* c t | S r, )r r- s r( r1 zFlyingThings3D._read_flow rr r) rs rM s @r( r r = s D )-'51:CI 1: 1: 1:
1: X&
1: #$1:
1:f
*
*r2v
* $C $BJJ $r) r c e Zd ZdZdZddefdeeef dede e
de
ege f d df
fd
Z d ed e
ej ej f fd Zd
ed eeef f fd Z xZS )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.
TrO Nr rQ r r r c
t
| ||| t |dd t | dz }|dk( rt d D ] }t t
t |dz d z |d
d z }t t
t |d z d
z |d
d z }t t | dz
D ]< }| xj || gz
c_ | xj || ||dz ggz
c_
> nxt t
t |dz d
z dz } t t
t |dz d
z dz }
t | |
D ] \ } } | xj | | ggz
c_
| j s t d y )NrT rQ rU rW hd1krO $ hd1k_flow_gtr{ 06dz_*.png
hd1k_inputrz r hd1k_challengez*10.pngz*11.pngzTCould not find the HD1K images. Please make sure the directory structure is correct.)
r! r" r r rb ra r rG r; r# r$ r} r| )r& r rQ r r seq_idxr r ri r~ r image1image2r' s r( r" z
HD1K.__init__ s dz&Iug4EFDzF" G 9
E tC~(=
(JPWX[}\bMc(c$d ef S )DC DE"**bjj2H,I D
*
*r2v
*
*r) r r* r c t | d 5 }t j |dd j }|dk7 r t d t j |dd j }t j |dd j }t j |d d
|z |z }|j
||d
j d
d d cd d d S # 1 sw Y y xY w)
z#Read .flo file in Middlebury formatrbc )counts PIEHz)Magic number incorrect. Invalid .flo filezr s # 1 1 , " , !
5;; Xbjj%98BJJ;O OP
5;; Xbjj%9 9: 7:#} 7:tX$[ X$vCD CDLA$; A$Hc$[ c$LI*; I*X8 8 8"1 1rzzSUS]S]G]A^ 1r)