Fi
d d dl mZ d dlZd dlmZ d dlmZ d dlmZ d dl Z d dl
Z d dl Z d dl
mZmZmZ d dl
mZ d dlmZmZmZmZmZmZmZ d d lm Z m Z m Z d d
l m Z G d d Z! e! Z" G d
d Z# e e d ed dfd Z$ ed dddddf d%dZ%e e jL de jN dddddd dd f d&d Z( e d'd(d Z)d)d*d Z* e d+d,d! Z+ e d" ed# fd-d$ Z,y). ) annotationsN)Callable)Path)Any)Image ImageDraw ImageFont) __version__)IS_COLAB IS_KAGGLELOGGER TryExceptops plt_settingsthreaded)
check_font
check_versionis_ascii)increment_pathc 2 e Zd ZdZd ZdddZedd Zy) Colorsa Ultralytics color palette for visualization and plotting.
This class provides methods to work with the Ultralytics color palette, including converting hex color codes to RGB
values and accessing predefined color schemes for object detection and pose estimation.
## Ultralytics Color Palette
| Index | Color | HEX | RGB |
|-------|-------------------------------------------------------------------|-----------|-------------------|
| 0 | | `#042aff` | (4, 42, 255) |
| 1 | | `#0bdbeb` | (11, 219, 235) |
| 2 | | `#f3f3f3` | (243, 243, 243) |
| 3 | | `#00dfb7` | (0, 223, 183) |
| 4 | | `#111f68` | (17, 31, 104) |
| 5 | | `#ff6fdd` | (255, 111, 221) |
| 6 | | `#ff444f` | (255, 68, 79) |
| 7 | | `#cced00` | (204, 237, 0) |
| 8 | | `#00f344` | (0, 243, 68) |
| 9 | | `#bd00ff` | (189, 0, 255) |
| 10 | | `#00b4ff` | (0, 180, 255) |
| 11 | | `#dd00ba` | (221, 0, 186) |
| 12 | | `#00ffff` | (0, 255, 255) |
| 13 | | `#26c000` | (38, 192, 0) |
| 14 | | `#01ffb3` | (1, 255, 179) |
| 15 | | `#7d24ff` | (125, 36, 255) |
| 16 | | `#7b0068` | (123, 0, 104) |
| 17 | | `#ff1b6c` | (255, 27, 108) |
| 18 | | `#fc6d2f` | (252, 109, 47) |
| 19 | | `#a2ff0b` | (162, 255, 11) |
## Pose Color Palette
| Index | Color | HEX | RGB |
|-------|-------------------------------------------------------------------|-----------|-------------------|
| 0 | | `#ff8000` | (255, 128, 0) |
| 1 | | `#ff9933` | (255, 153, 51) |
| 2 | | `#ffb266` | (255, 178, 102) |
| 3 | | `#e6e600` | (230, 230, 0) |
| 4 | | `#ff99ff` | (255, 153, 255) |
| 5 | | `#99ccff` | (153, 204, 255) |
| 6 | | `#ff66ff` | (255, 102, 255) |
| 7 | | `#ff33ff` | (255, 51, 255) |
| 8 | | `#66b2ff` | (102, 178, 255) |
| 9 | | `#3399ff` | (51, 153, 255) |
| 10 | | `#ff9999` | (255, 153, 153) |
| 11 | | `#ff6666` | (255, 102, 102) |
| 12 | | `#ff3333` | (255, 51, 51) |
| 13 | | `#99ff99` | (153, 255, 153) |
| 14 | | `#66ff66` | (102, 255, 102) |
| 15 | | `#33ff33` | (51, 255, 51) |
| 16 | | `#00ff00` | (0, 255, 0) |
| 17 | | `#0000ff` | (0, 0, 255) |
| 18 | | `#ff0000` | (255, 0, 0) |
| 19 | | `#ffffff` | (255, 255, 255) |
!!! note "Ultralytics Brand Colors"
For Ultralytics brand colors see [https://www.ultralytics.com/brand](https://www.ultralytics.com/brand).
Please use the official Ultralytics colors for all marketing materials.
Attributes:
palette (list[tuple]): List of RGB color tuples for general use.
n (int): The number of colors in the palette.
pose_palette (np.ndarray): A specific color palette array for pose estimation with dtype np.uint8.
Examples:
>>> from ultralytics.utils.plotting import Colors
>>> colors = Colors()
>>> colors(5, True) # Returns BGR format: (221, 111, 255)
>>> colors(5, False) # Returns RGB format: (255, 111, 221)
c ` d}|D cg c] }| j d| c}| _ t | j | _ t j
g dg dg dg dg dg dg d g d
g d g d g d
g dg dg dg dg dg dg dg dg dgt j | _ yc c}w )zEInitialize colors as hex = matplotlib.colors.TABLEAU_COLORS.values().)042AFF0BDBEBF3F3F300DFB7111F68FF6FDDFF444FCCED0000F344BD00FF00B4FFDD00BA00FFFF26C00001FFB37D24FF7B0068FF1B6CFC6D2FA2FF0B#) r )r. 3 )r. f ) r4 r )r. r0 r. )r0 r. )r. r3 r. )r. r1 r. )r3 r2 r. )r1 r0 r. )r. r0 r0 )r. r3 r3 )r. r1 r1 )r0 r. r0 )r3 r. r3 )r1 r. r1 )r r. r )r r r. )r. r r r. r. r. dtypeN)hex2rgbpalettelennnparrayuint8 pose_palette)selfhexscs \/var/www/pod-logistic/pod-ai/venv/lib/python3.12/site-packages/ultralytics/utils/plotting.py__init__zColors.__init__^ s
, 8<
LLQ$&& )%(!adAaD!/a/ c , t fddD S )z?Convert hex color codes to RGB values (i.e. default PIL order).c 3 N K | ] }t d |z d |z dz d yw)rH rG NrI ).0rJ hs rD z!Colors.hex2rgb.. s+ FqS1q51q519-r2Fs "%)r rG )tuple)rS s `rD r9 zColors.hex2rgb s FIFFFrM NF)rJ zint | torch.TensorrK boolreturnrV )rS strrY rV )__name__
__module__ __qualname____doc__rE rL staticmethodr9 rM rD r r s* FP2
h 0 G GrM r c e Zd ZdZ d ddZdddZdddZdddZ d ddZdddZ dd d Z
d
Z d d Z d d d Z
d d!d
Zed"d Zy)# Annotatora Ultralytics Annotator for train/val mosaics and JPGs and predictions annotations.
Attributes:
im (Image.Image | np.ndarray): The image to annotate.
pil (bool): Whether to use PIL or cv2 for drawing annotations.
font (ImageFont.truetype | ImageFont.load_default): Font used for text annotations.
lw (int): Line width for drawing.
skeleton (list[list[int]]): Skeleton structure for keypoints.
limb_color (np.ndarray): Color palette for limbs.
kpt_color (np.ndarray): Color palette for keypoints.
dark_colors (set): Set of colors considered dark for text contrast.
light_colors (set): Set of colors considered light for text contrast.
Examples:
>>> from ultralytics.utils.plotting import Annotator
>>> im0 = cv2.imread("test.png")
>>> annotator = Annotator(im0, line_width=10)
>>> annotator.box_label([10, 10, 100, 100], "person", (255, 0, 0))
Nc t | }t |t j }|xs |xs | _ |xs< t t t
|r |j n |j dz dz d _ |s|j d dk( r%t j |t j }n|j d dk( rEt j t j |t j |dddf f }n.|j d dkD r t j |dddf } j r|r|nt j" | _ j$ j& dvr j$ j) d _ t+ j, j$ d _ t1 |rd
n| }|xs8 t t t
j$ j dz d z d } t3 j4 t7 | | _ t? t@ d
r fd j8 _! n||jD jF sJ d |jH jJ r|n|jM _ t j dz
d _' j dz _( ddgdd gddgddgd dgdd gddgddgddgddgddgddgddgddgddgddgdd gddgd dgg _) tT jV g d _, tT jV g d _- h d _. h d _/ y# t: $ r t3 j< _ Y Mw xY w)!zjInitialize the Annotator class with image and line width along with color palette for keypoints and limbs.rG g~jth?rH .N > RGBRGBAre rf zArial.Unicode.ttfgQ ? z9.2.0c @ j j | dd S )NrG rU )fontgetbbox)xrA s rD z$Annotator.__init__.. s dii.?.?.B1Q.G rM zOImage not contiguous. Apply np.ascontiguousarray(im) to Annotator input images.rP
rU )rt rt rt rt rr rr rr r r r r r rP rP rP rP rP rP rP )rP rP rP rP rP r r r r r r rt rt rt rt rt rt > r r5 rv r. D r r. rH r o r. rv r~ r~ r~ r. r. r > r & / m O r} r. h r { l r. r r r. r r. $ } r. * rU r. r r rn )0r
isinstancer pilmaxroundsumsizeshapelwcv2cvtColorCOLOR_GRAY2BGRr= ascontiguousarraydstack
zeros_like fromarrayimmodeconvertr Drawdrawr r truetyperZ ri Exception load_defaultr pil_versiongetsizedata
contiguousflags writeablecopytfsfskeletoncolorsr@
limb_color kpt_color dark_colors light_colors)
rA r
line_width font_sizeri r example non_ascii input_is_pilr s
` rD rE zAnnotator.__init__ s !)) !"ekk2 3)3| eE# bgg288*TWX*X[`*`$acd exx{a \\"c&8&89!!))"))R r#rPQr'{9S4T*UV!q ))"S"1"W+6 88 (beoob.ADGww||?2 ''//%0!tww7DI
5!"5M QCc$'',,.?!.Ce.K(Lb$Q % . .s4y$ ? ['2$G !77%% x'x x% HH . .bBGGIDG$''A+q)DGggkDG H
H
H
H
H
G
G
F
F
F
G
G
F
F
F
F
F
F
F'
, !--.no ,,-ef
]
5 % 2 2 4
5s A/M "N ?N c B || j v ry|| j v ry|S )a1 Assign text color based on background color.
Args:
color (tuple, optional): The background color of the rectangle for text.
txt_color (tuple, optional): The fallback color of the text.
Returns:
(tuple): Text color for label.
Examples:
>>> from ultralytics.utils.plotting import Annotator
>>> im0 = cv2.imread("test.png")
>>> annotator = Annotator(im0, line_width=10)
>>> annotator.get_txt_color(color=(104, 31, 17)) # return (255, 255, 255)
r r6 )r r )rA color txt_colors rD
get_txt_colorzAnnotator.get_txt_color s, D$$ $
d''
' rM c
| j || }t |t j r|j }t |d t
}|r |d D cg c]
}t
| c}nt
|d t
|d f}| j r^|r?| j j |D cg c]
}t | c}| j | n'| j j || j | |r| j j | \ }} |d | k\ }
|d | j j d |z
kD r!| j j d |z
|d f}| j j |d |
r|d | z
n|d |d |z dz |
r|d dz n
|d | z dz f| | j j# |d |
r|d | z
n|d f||| j y
y
|rGt% j& | j t) j* |t gd|| j nVt% j | j |t
|d t
|d f|| j t$ j,
|r1t% j. |d| j0 | j2 d \ }} | d z
} |d | k\ }
|d | j j4 d |z
kD r!| j j4 d |z
|d f}|d |z |
r|d | z
n|d | z f} t% j | j || |d t$ j, t% j6 | j ||d |
r|d dz
n
|d | z dz
fd| j0 || j2 t$ j,
y
y
c c}w c c}w )ag Draw a bounding box on an image with a given label.
Args:
box (tuple): The bounding box coordinates (x1, y1, x2, y2).
label (str, optional): The text label to be displayed.
color (tuple, optional): The background color of the rectangle.
txt_color (tuple, optional): The color of the text.
Examples:
>>> from ultralytics.utils.plotting import Annotator
>>> im0 = cv2.imread("test.png")
>>> annotator = Annotator(im0, line_width=10)
>>> annotator.box_label(box=[10, 20, 30, 40], label="person")
r rH widthoutlinefillr ri r7 TrG rd thicknesslineType fontScaler N) r r torchTensortolistlistrI r r polygonrV r rectangleri r r r textr polylinesr= asarrayLINE_AA getTextSizer r r putText) rA boxlabelr r multi_pointsbp1wrS outsidep2s rD box_labelzAnnotator.box_label. ss &&ui8 c5<< (**,C!#a&$/ )5c!f
%c!f
%CAKSQRV;U 88
II #&'aq'tww
#'99#6#6s$''SX#6#Yyy((/1 Q%1*a5477<<?Q.. a 1,be3B ##UBqEAIbeRUQY]Y`BqETUIfhijfknoforsfst $
1Gr!uqyA GU^eieneno
MM"**S45tUDGG
#&==c#a&k3s1v; 7$''\_\g\g$ ua477dggVWXY1Q Q%1*a5477==+a// q)A-r!u4BUQYW1 "Q%!)C
dggr2ub#++F GGUBqEAIbeai!mDGG "gg [[ - & (s O
2Oc | j r2t j | j j | _ |t |t j sJ d | j j }t | D ] \ }}|| ||j t < " t j | j d|z
||d | _ nwt |t j sJ d t | dk( rN|j ddd j! j# j% dz | j dd y|j& |j& k7 r|j) |j& }| j j* dd \ } }
|st- j. |d j1 | |
f d dkD }t j2 | j j) |j& j ddd j5 d j! j1 d z }t j6 ||j& t j8
d z }|ddddf }|j; d }|||z z } d||z z
j= d } | j? d j@ }
|j5 dg
j ddd j! }|| d z |
z }|dz jC j# j% | j dd | j r | jE | j yy)a Plot masks on image.
Args:
masks (torch.Tensor | np.ndarray): Predicted masks with shape [n, h, w].
colors (list[list[int]]): Colors for predicted masks, [[r, g, b] * n].
im_gpu (torch.Tensor | None): Image on GPU with shape [3, h, w], range [0, 1].
alpha (float, optional): Mask transparency: 0.0 fully transparent, 1.0 opaque.
retina_masks (bool, optional): Whether to use high resolution masks or not.
Nz9`masks` must be a np.ndarray if `im_gpu` is not provided.rH r z7'masks' must be a torch.Tensor if 'im_gpu' is provided.rG r. ?g o@)devicer8 rd dim)dimsr )#r r= r r r r ndarray enumerateastyperX r addWeightedr r r; permuter cpunumpyr tor r scale_masksfloat
from_numpyfliptensorfloat32 unsqueezecumprodr valuesbyter )rA masksr im_gpualpha retina_masksoverlayrJ maskihiw masks_colorinv_alpha_masksmcss rD r zAnnotator.masksk s 88jj)..0DG > eRZZ0 m2m m0ggllnG$U+
74-3AY D)*
7oodggq5y'5!LDG eU\\2 m4m m25zQ #^^Aq! 4 ? ? A E E G M M ORU U
}} , 5<<0WW]]2A&FB d (9(9(;b"XFqICO $$TWW-00 >FFq!QOTTUVWbbdjjlott \\& U]]SV[[FAtTM*FOOA&E 6E>2K 55= 0 9 9! rG rd r rH rd rG r )r r )r r r= r r r r r r r r r; r circlerI r r liner ceilr )rA kptsr radiuskpt_line
conf_thresr nkptndimis_poserJ kcolor_kx_coordy_coordconfskpos1pos2conf1conf2s rD r zAnnotator.kpts s 0 "-477 88jj)..0DGZZ
d"*/ G dO mDAq Y'DNN1$5$<$<$>vVWyG tQqTWGq!Q&7U1X+=+Bq6Q; Q4Dj(
477S\3w<$@&'SU`c`k`kl m ::b>D"4==1
2D"Q%!)a 013tRUQYN7K3LMD"Q%!)a 013tRUQYN7K3LM19 "Q%!)a 0E "Q%!)a 0E z)UZ-? 7U1X%*d1ga.@A.EaSTX\]^X_bcXc 7U1X%*d1ga.@A.EaSTX\]^X_bcXc GG < !3!:!:!< !"''$''A+"6 7 [[
( 88 NN477 # rM c @ | j j |||| y)z"Add rectangle to image (PIL-only).N)r r )rA xyr r r s rD r zAnnotator.rectangle s Bgu5rM c
| | j r| j j | \ }}|dk( r|dxx d|z
z
cc< |j d D ] }|rW| j j | \ }}| j j |d |d |d |z dz |d |z dz f| | j j
|||| j |dxx |z
cc< y |rt j |d| j | j d \ }}|dz
}|d |k\ } |d |z | r|d |z
n|d |z f}
t j
| j ||
|d t j t j | j ||d| j || j t j
y ) a Add text to an image using PIL or cv2.
Args:
xy (list[int]): Top-left coordinates for text placement.
text (str): Text to be drawn.
txt_color (tuple, optional): Text color.
anchor (str, optional): Text anchor position ('top' or 'bottom').
box_color (tuple, optional): Box background color with optional alpha.
bottomrH
r r r r rd r r N)r ri r splitr r r r r r r r r r ) rA r2 r r anchor box_colorr rS r r r s rD r zAnnotator.text s 8899$$T*DAq!1Q
4(
99,,T2DAqII''A1r!uqy1}beaiRSm(T[d'e r4idiiH1
tQ$''TWWUVWX1Q Q%1*UQYW1 "Q%!)C
dggr2y"ckkJ KKr1dggyDGG^a^i^i jrM c t |t j r|nt j | | _ t j
| j | _ y)z1Update `self.im` from a NumPy array or PIL image.N)r r r r r r r )rA r s rD r zAnnotator.fromarray s5 "2u{{3" 9L NN477+ rM c t j | j }|r t j |ddddf S |S )z-Return annotated image as array or PIL image..Nr )r= r r r r )rA r r s rD resultzAnnotator.result s6
ZZ
14ur#tt)}-<">> from ultralytics.utils.plotting import Annotator
>>> im0 = cv2.imread("test.png")
>>> annotator = Annotator(im0, line_width=10)
>>> annotator.get_bbox_dimension(bbox=[10, 20, 30, 40])
r` )bboxx_miny_minx_maxy_maxr heights rD get_bbox_dimensionz Annotator.get_bbox_dimension s5 $ &*"ueU
fefn,,rM )NNz Arial.ttfFabc)
r
int | Noner rP ri rZ r rX r rZ )r/ r/ r/ r6 )r rV r rV rY rV ) rQ r6 )r rZ r rV r rV )Nr F)r z torch.Tensorr r r rX )) rS NT ?N)
r rV r! rP r" rX r# r r z tuple | None)NNrH )r rI )r6 topr` )r rZ r rV r7 rZ r8 rV rW N)r= z
str | None)z image.jpg)rE rZ )rH z tuple | list)r[ r\ r] r^ rE r r r r r r r r; rA rF r_ rN r` rM rD rb rb s . "& $ X
X
X
X
X
X
t .;z0$j " ! "&A$ A$ A$
A$
A$ A$F6 k<,
=
!3 - -rM rb r` rR c X ddl m} ddl}ddlm} t j d|dz d t |j dz }| dd } |j | g d
} |j d d d
g }
|j dddd d j } | d j |t j d||dz dz
d } t! | D ]= }
| d j" |
j% t' |
D cg c] } | dz c} ? | d j) d dt+ | cxk rdk ron nl| d j- t! t+ | | d j/ t1 |j3 dd | d j5 | d n| d j7 d t j8 d| dddd f dz z
d| dddd f dz z g d z } t; j< t j> d t j@ dz }tC |dd | dd D ]C \ }}tE jF | jI |jK dt' | ! E | d jM | | d jO d" | d jQ d# | d$ d%|
& | d j7 d# | d j) d$ | d' jQ | d( | d) d%|
& | d' j7 d( | d' j) d) d*D ]* }d+D ]# }| | jR | jU d, % , |dz }|jW |d-. |jY |r || yyc c} w )/a Plot training labels including class histograms and box statistics.
Args:
boxes (np.ndarray): Bounding box coordinates in format [x, y, width, height].
cls (np.ndarray): Class indices.
names (dict, optional): Dictionary mapping class indices to class names.
save_dir (Path, optional): Directory to save the plot.
on_plot (Callable, optional): Function to call after plot is saved.
r N)LinearSegmentedColormapzPlotting labels to z
labels.jpgz... rH i@B )rk yr rM )schema
white_bluewhitebluerG )rs rs Tfigsize tight_layoutr 皙?)binsrwidthr. instances Z ru )rotationfontsizeclassesrU )rj rj rd r7 i r offrk rY 2 )rb cmaprd r rM > r rH rG rd > rU leftrightr4 F dpi)-matplotlib.pyplotpyplotpolarsmatplotlib.colorsrX r
inforI r DataFrame from_listsubplotsravelhistr= linspacerangepatches set_colorr
set_ylabelr;
set_xticksset_xticklabelsr r bar_label
set_xlabel column_stackr r onesr? zipr r r r imshowaxishist2dspines set_visiblesavefigclose)boxesclsnamessave_diron_plotpltru rX ncrk subplot_3_4_coloraxrY rJ imgclass_idr asfnames rD plot_labelsr 3 sj $9 KK%h &=%>dCD SWWY] B (7OE'DEA 099,RXHYZ aF >q A G G IB
1
3R[[BQ 7# =c
JA
2Y @ ! Q!!F1I">q1s7">?@qE[!3u:
1s5z*+
1 d5<<> 2R" M
1!
1# OOS5AaC=1#4 4cE!QqS&MA>> from ultralytics.utils.plotting import save_one_box
>>> xyxy = [50, 50, 150, 150]
>>> im = cv2.imread("image.jpg")
>>> cropped_im = save_one_box(xyxy, im, file="cropped.jpg", square=True)
r rU NrG rH r )r rH )r rd )r r )r rG T)parentsexist_okz.jpg._ )quality subsampling)r r r stackr xyxy2xywhviewr r
xywh2xyxylong
clip_boxesr rI parentmkdirrZ r with_suffixsqueezer r rF ) xyxyr filegainpadsquareBGRrF r grayscalecropfs rD save_one_boxr n s F dELL ){{4
diiA&'A
QU8<<?1%//2!QR%ABx$$AaeH == "D >>$ )D q I
c$t*oDJ/T$Z 3tDz?1RTv]`dmXYsuTvv wD $6 t$008 9#,t||B S$sDbDy/d
""1ba"@ KrM )r rd rS rS r7 z
images.jpgi rP rT c
/ dD ]w }
|
| vr|
dk( r)| |
j dk( r| |
j d | |
< t | |
t j sT| |
j j
| |
< y | j dt j dt j } | j dt j | j t j } | j dt j dt j }
| j d d
}| j d t j dt j }| j d t j dt j }| j d
| }t | rFt |t j r,|j j j
}|j d }|dk( r8t j |d
d
d
df }t j" ||fd }n|dkD r |d
d
d
df }|j \ }}}}t% || }t j& |dz /t j( |d dk r|dz }t j* t- /|z t- /|z dfdt j }t/ | D ]L }t- ||/z z t- ||/z z }}|| j1 ddd ||||z |||z d
d
f< N |/z t) || z }|dk rZt3 j& ||z }t3 j& ||z }t5 j6 |t9 /fd||fD }t- ||z /z dz } t) | d } t; |t= | dz | dt? | } t/ | D ] }t- ||/z z t- ||/z z }}| jA ||||z ||z gd
dd |r5| jC |dz |dz gtE || jF d
d d t | dkD s| |k( } | | jI d } |d
u } ||| nd
} t |
rQ|
| }!t |! rU|!d
d
d
d f j) d k r#|!d!ddgfxx |z cc< |!d!ddgfxx |z cc< n|dk r|!d!d
d fxx |z cc< |!d"xx |z
cc< |!d#xx |z
cc< |!j d$ dk( }"|"rtK jL |! ntK jN |! }!tQ |!jI t j jS D ]Z \ }#}$| |# }tU | }%|r|j || n|}| s | |# | kD s7| r| n
| d%| |# d&}&| jW |$|&|%' \ nZt | rO| D ]J }tU | }%|r|j || n|}| r| n
| d%| d d&}&| jC ||g|&|%d() L t | r|| jY }'t |' rQ|'d" j) d*k s|'d# j) d*k r|'d"xx |z cc< |'d#xx |z cc< n
|dk r|'|z }'|'d"xx |z
cc< |'d#xx |z
cc< t/ t |' D ]# }#| s | |# | kD s| j[ |'|# | + % t | sC| j d |j d k( r|j) dk r|| }(nc||g }(| j] })t j^ d|)dz ja |)ddf }*|(|*k( jI t j }(t jb | jd jY }+t/ t |( D ] }#| s | |# | kD stU | |# }%|(|# j \ },}-|,|k7 s|-|k7 rP|(|# jI t j }.t5 j6 |.||f }.|.jI tf }.n|(|# jI tf }. |+|||z |||z d
d
f |. d,z t jh |% d-z z |+|||z |||z d
d
f |.< | jm |+ |s t jb | jd S | jd jo | |r || y
y
# tj $ r Y Sw xY w).a Plot image grid with labels, bounding boxes, masks, and keypoints.
Args:
labels (dict[str, Any]): Dictionary containing detection data with keys like 'cls', 'bboxes', 'conf', 'masks',
'keypoints', 'batch_idx', 'img'.
images (torch.Tensor | np.ndarray): Batch of images to plot. Shape: (batch_size, channels, height, width).
paths (list[str] | None): List of file paths for each image in the batch.
fname (str): Output filename for the plotted image grid.
names (dict[int, str] | None): Dictionary mapping class indices to class names.
on_plot (Callable | None): Callback function to be called after saving the plot.
max_size (int): Maximum size of the output image grid.
max_subplots (int): Maximum number of subplots in the image grid.
save (bool): Whether to save the plotted image grid to a file.
conf_thres (float): Confidence threshold for displaying detections.
Returns:
(np.ndarray | None): Plotted image grid as a numpy array if save is False, None otherwise.
Notes:
This function supports both tensor and numpy array inputs. It will automatically
convert tensor inputs to numpy arrays for processing.
Channel Support:
- 1 channel: Grayscale
- 2 channels: Third channel added as zeros
- 3 channels: Used as-is (standard RGB)
- 4+ channels: Cropped to first 3 channels
> r r+ r bboxesimages batch_idx keypointsr rG rH r r7 r r r+ Nr r r )r rd r r. c 3 : K | ] }t |z y wrV rQ )rR rk nss rD rT z plot_images.. s )F!#a"f+)Fs g{Gz? ru T)r r r r r6 )r rw ( ) r r )r r rI rU g?.).r ).rH r z.1f)r )@ r r r/ )r r8 g)\(?)r# g?g333333?)8r% r r r r r r getr= zerosint64r r r? r; r r concatenateminr r fullrI r~ transposemathr resizerV rb r rZ r r r namer r xywhr2xyxyxyxyr r r r r r r r arangereshaper r rX r> r r rF )0labelsr pathsr r r max_size max_subplotsrF r# r' r r r confsr r rC zerobs_rS r mosaicrJ rk rY scalefs annotatoridxri r+ r is_obbjr r r kpts_ image_masksnlindexr mhmwr r s0 @rD plot_imagesr s R T 0 F? :&)..A- q ))!,F1I fQi . q
--/F1I
0 **UBHHQbhh 7
8C
;"(((KLI
ZZ"((1BJJ"?
@F JJvt $E JJw"(( ; U1X0C0CCR0HTcN d s8a<q.C#hooe,Gd]F!&!25:D6{ s u: QU|'')S0 cAq6k*a/* cAq6k*a/* c2A2g%/f
"
f
"
R A-5;**51 uAU ' RXX(>(E(E(G H EFAs
A "1IE+0 !QaA a:!5*01#1T!WSM6J!++Ce+D
E W `A "1IE+0 !QaA&,qcQCqa
2FE NNAq65EM^N_ ` 4y S (u: V}((*d2eFm6G6G6IT6Q f
*
f
*
f
"
f
"
s5z* HA a:!5!uQxJGH
5z99Q<5;;q>1eiikQ6F"'*K"'*K B IIaa 0 8 8"a DE#.%#7"?"?
"KKZZ -224 s; /0 !A a:!5 &wqz 2!,Q !5!5B 7bAg#.q>#8#8#BD#&::dQF#;D#';;t#4D#.q>#8#8#>D! "1q1u9a!a%i#: ;D AC G"((SX/\_J_ _ q1q5y!a!e)Q 6 7 =! ##B'aP(b zz),,''
LLe
) ! !s Ae99 ffc
t ddl m} ddl}ddlm} | rt | j n
t | }t |j d }t | sJ d|j d g g } }d\ }
} t | D ] \ } }
|j |
d }| dk( r|j D ]. }d |v r|j | d
|v s | j | 0 t | d z t | d z }}|d| | d| z ||d z | |d z }|j d t | d z t | d z d fd
\ }
} | j! } |j# |j d j% j' }t D ] \ } }|j# | j% j' j) d }| | j+ ||d|
j, d d | | j+ | ||d ddd | | j/ |d | G| d j7 |d z }|
j9 |d |j; |r || yyy# t0 $ r&}t3 j4 d|
d| Y d}~-d}~ww xY w) a Plot training results from a results CSV file. The function supports various types of data including
segmentation, pose estimation, and classification. Plots are saved as 'results.png' in the directory where the
CSV is located.
Args:
file (str, optional): Path to the CSV file containing the training results.
dir (str, optional): Directory where the CSV file is located if 'file' is not provided.
on_plot (Callable, optional): Callback function to be executed after plotting. Takes filename as an argument.
Examples:
>>> from ultralytics.utils.plotting import plot_results
>>> plot_results("path/to/results.csv")
r Ngaussian_filter1dz results*.csvz No results.csv files found in z, nothing to plot.)NN)infer_schema_lengthlossmetricrG rq Tr^ r .rs )markerr linewidth
markersizerd sigma:smooth)r r rg )rh zPlotting error for z: rH z results.pngrp rq ) rs rt ru
scipy.ndimager r r r globr; resolver read_csvcolumnsappendrz r{ selectto_numpyflattenr plotstem set_titler r
errorlegendr r )r dirr r plr r files loss_keys metric_keysfigr rJ r r rC loss_mid
metric_midr rk r rY rB r s rD plot_resultsr W s $/$(tDz d3iH ~. /E u:^78H8H8J7KK]^^: {IGC% 91 9;;qd;;DAv .A {!((+!Q #**1- .
(+9~':C >wG1
1a166QST
U1
1 / ;S\]
^1B/
0!9.
~
1 =( Es # EN 9 LL.qcA37 8 8 9s
AJE,J J7J22J7c ddl m} t j | || \ }}} t t | D
cg c]u }
|t
t j | |
|d dz
|j d dz
t
t j ||
| d dz
|j d dz
f w } }
|j | || ||| yc c}
w )a) Plot a scatter plot with points colored based on a 2D histogram.
Args:
v (array-like): Values for the x-axis.
f (array-like): Values for the y-axis.
bins (int, optional): Number of bins for the histogram.
cmap (str, optional): Colormap for the scatter plot.
alpha (float, optional): Alpha for the scatter plot.
edgecolors (str, optional): Edge colors for the scatter plot.
Examples:
>>> v = np.random.rand(100)
>>> f = np.random.rand(100)
>>> plt_color_scatter(v, f)
r N)rb T)ro rH )rC rm r
edgecolors)
rs rt r= histogram2dr~ r; r digitizer scatter) vr rb rm r r r r| xedgesyedgesrJ r s rD plt_color_scatterr s $ >>!QT:D&& s1v
AaD&594::a=1;L M AaD&594::a=1;L M
O
F KK1T:KNs A:C
c ddl }ddlm ddlm} fd}t | } t
| d 5 }|D cg c]$ }|j s|j | & }}ddd syt |d j di }t j |D
cg c]T } | j dd g|D
cg c]2 }
| j di j |
t j 4 c}
z V c}
} t
} t | | dddf } t! d }
|r| dkD }
| |
| |
} } t | dk( rt# j$ d y| j' }t) d D ]K }|j+ |j- }}|d |z z
}||k\ }|j/ r n | | || }} M t j0 | }t3 j4 t | d
z }j7 dd t9 | D ] \ }}
| dd|dz f }|| }j; |||dz t= ||ddd j? ||jA dd jC |
d|ddd i jE d d ||z dk7 sjG g || jI d! t) dt | dz } j7 d"d tK |D
ch c] } | j d#i D ] }
|
c}
} D ] }t j |D cg c]B } | j d#i j |i j dt j D c} t
}|rtM |
t s||
}j? | |d$d%d|& j? | || d ' d(d)d*d+, jC d- jO d. jQ d/ jS d jU || jI d0 yc c}w # 1 sw Y xY wc c}
w c c}
} w c c}
} w c c} w )1aa Plot the evolution results stored in a tuning NDJSON file.
Args:
results_file (str, optional): Path to the NDJSON file containing the tuning results.
exclude_zero_fitness_points (bool, optional): Don't include points with zero fitness in tuning plots.
Examples:
>>> plot_tune_results("path/to/tune_results.ndjson")
r Nr c | j | d j t j d| y)z#Save one matplotlib plot to 'file'.rp rq zSaved N)r r r
rw )r r s rD _save_one_filez)plot_tune_results.._save_one_file s. Dc " fTFO$rM zutf-8)encodinghyperparametersfitnessg r7 z@No valid fitness values to plot (all iterations may have failed)rd r )ru ru Tr^ rH viridisra none)rm r r zk+ro )r z = z.3gr rt )fontdictbothrs )r labelsizeztune_scatter_plots.png)ru rq datasetsorw )r r r r r z0.35z
smoothed meanrG )r r r zFitness vs Iteration IterationFitnessztune_fitness.png)+jsonrs rt r r r openstriploadsr r r= r> nanr r; slicer
r@ r r~ meanstdallargmaxr r figurer subplotr r r r= tick_paramsyticks with_namesortedr xlabelylabelgridr ) results_fileexclude_zero_fitness_pointsr% r r r r recordskeysrr' rk all_fitness zero_maskr r r+ r, lower_boundr r r< rJ r mudatasetrY r s @rD plot_tune_resultsrB s
#/% %L
lW - C01BTZZ\4::d#BBC
0"5 6D
hopcd!%% 3
Y]#^TUAEE*;R$@$D$DQ$O#^ ^p A FAqD'Kd I" !O 9 {9'=;
;1 YZ G
1X ,LLNGKKMcQWn +% 88: tWgdm7
, 'A #d)s"#AJJxdJ3$ 1
aQhK
qT Aq!a% !W9CFS W[[]DR8 QCs2c(#vqk : Vq1 q5A: JJrN <))*BCD a[!A%&AJJwTJ2gMquuZ7LM!1M1MN D HH]deXYaeeJ+//<@@BFFSemr s &z)U/K) A Asq7C D
HHQ!+Q7FRamnHoII$%JJ{JJyHHTNJJL<))*<=>q CC C $_pN NesG Q Q
Q Q Q.
77Q).Q.
Q4
AQ:
Q Q&)Q.
zruns/detect/expc ddl m} dD ] }||v s y t | t j r| j
\ }}} }
| dkD r|
dkD r~|d| d|j
dd d d z } t j | d j |d
} t || }|j t j |d z d d
\ }}
|
j }
|j dd t | D ]; }|
| j! | | j# |
| j% d = t' j( d| d| d| d |j+ | dd |j- t/ j0 t3 | j5 d | d j j7 yyyy)ab Visualize feature maps of a given model module during inference.
Args:
x (torch.Tensor): Features to be visualized.
module_type (str): Module type.
stage (int): Module stage within the model.
n (int, optional): Maximum number of feature maps to plot.
save_dir (Path, optional): Directory to save results.
r N> OBBPoseDetectSegmentClassify
RTDETRDecoderrH stager r r z
_features.pngr rs T)r` g?)wspacehspacerk zSaving z... (/)i, tight)rr bbox_inchesz.npy) rs rt r r r r rsplitchunkr r rz r r r{ subplots_adjustr~ r r r r
rw r r r= rF rZ r r )rk module_typerK r< r r mr channelsrM r r blocksr rJ s rD feature_visualizationrY s $
N !U\\"%&WW"8VU A:%!)U5';+=+=c1+Eb+I*J-XXA[[1X1=FAx ALL1q5!114LHEArB tD 91X
"1 VAY . . 011
5!
"
KK'!E!AhZq9 : KKsK 8 IIK GGC
f-.!
0@0@0B C $: #rM ) r r r r r rI r rX r rX rF rX )r zdict[str, Any]r ztorch.Tensor | np.ndarrayr zlist[str] | Noner rZ r zdict[int, str] | Noner Callable | Noner rI r rI rF rX r# r rY znp.ndarray | None)zpath/to/results.csvrR N)r rZ r rZ r rZ ) r ra r )rb rI rm rZ r r r rZ )ztune_results.ndjsonT)r8 rZ r9 rX )rU rZ rK rI r< rI r r )-
__future__r r collections.abcr pathlibr typingr r r r= r PILr r r r
r ultralytics.utilsr r r
r r r r ultralytics.utils.checksr r r ultralytics.utils.filesr r r rb r r r r r r r rB rY r` rM rD rd s # $
+ + * a a a H H 2MG MG`
H- H-V
"$tBx 6
6x h 3 3 3
3
3
3 3l
)11A(T " #' # o o %o o o
! o
o o o o o o
od 5 5p O@ O? O?d DFY]^oYp D DrM