The document lists over 30 tools for annotating images, videos, and point cloud data. Many of the tools are open source and used for tasks like object detection, segmentation, and labeling. The tools cover a wide range of domains from natural images to LiDAR point clouds and include both online and desktop-based annotation solutions.
5. } https://github.com/kyamagu/js-segment-annotator
} Javascript image annotation tool based on image segmentation.
◦ Label image regions with mouse.
◦ Written in vanilla Javascript, with require.js dependency (packaged).
◦ Pure client-side implementation of image segmentation.
} A browser must support HTML canvas to use this tool.
6. } It provides an online annotation tool to build image databases
for computer vision research;
} http://labelme.csail.mit.edu/Release3.0/
7. } GUI for marking bounded
boxes of objects in images
for training neural network
Yolo v3 and v2;
} https://github.com/Alexey
AB/Yolo_mark.
8. } Macro plugin to label images for Detectnet / KITTI dataset;
https://alpslabel.wordpress.com/2017/01/26/alt/
Work for Windows
and Ubuntu!
9. } An image segmentation tool.
} https://alpslabel.wordpress.com/2017/03/28/alps-image-segmentation-tool-aims/
Only for Windows!
10. } This Fiji plugin is to quickly verify if all the labeling data is in right place, and error free.
Work for Windows
and Ubuntu!
11. } An image annotation tool to label images for bounding box object
detection and segmentation.
} https://itunes.apple.com/jp/app/rectlabel-labeling-images-for-object-detection/id1210181730?mt=12
An iMac App!
Key features:
Drawing bounding box, polygon,
and cubic bezier
1-click buttons make your labeling
work faster
Customize the label dialog to
combine with attributes
Settings for objects, attributes,
hotkeys, and labeling fast
Search images whose labels
include keywords
Layer order for overlapped boxes
Zoom in on a point
Quick zoom to existing boxes
Support the PASCAL VOC format
13. } Pixel-wise object annotation
} Zoom in/out
} Different brush sizes (circle shape)
} Line drawing
} Flood filling
} Different color types: background, object,
occluded object
} Different drawing modes: over all or only
over a specific color type (i.e., masked)
} A mask file (in .png format) is created for
each object separately
https://lear.inrialpes.fr/people/klaeser/software_image_annotation
15. } Sloth’s purpose is to provide a versatile tool for various labeling
tasks in the context of computer vision research;
} https://github.com/cvhciKIT/sloth;
16. } A free, online, interactive video annotation tool for computer vision
research that crowdsources work to Amazon's Mechanical Turk;
https://github.com/cvondrick/vatic
C Vondrick, D Patterson, D Ramanan. “Efficiently Scaling Up Crowdsourced Video Annotation”!
International Journal of Computer Vision (IJCV). June 2012.
18. } An electron app for building end to end Object Detection Models
from Images and Videos from Microsoft;
https://github.com/Microsoft/VoTT/
19. } It allows to annotate regions on the images, and to associate to the
selected regions labels from a predefined taxonomy.
} The application allows to choose whether annotate a single image,
or several images.
} The application has been built using cross-platform Qt framework.
http://www.ivl.disco.unimib.it/activities/imgann/
21. } LabelD is a quick and easy-to-use image annotation tool, built for
academics, data scientists, and software engineers to enable single
track or distributed image tagging.
} LabelD supports image annotation as well as image categorization.
} https://github.com/sweppner/labeld!
} Dependencies
◦ NodeJS
◦ NPM
◦ NPM module - express
◦ NPM module - body-parser
◦ MongoDB
22. } Imglab is a simple graphical tool for annotating
images with object bounding boxes and optionally
their part locations.
} Generally, use it when training an object detector
(e.g. a face detector) since it allows to easily create
the needed training dataset.
} https://github.com/davisking/dlib/tree/master/tools/imglab!
24. } Given reconstructions from stereo or laser data, annotate static 3D
scene elements with rough bounding primitives and develop a model
which transfers this info. into the image domain.
https://arxiv.org/abs/1511.03240
25. (CRF model)
Label Transfer Model. (a) Factor graph
representation of our model. (b) 3D structures
such as folds and curbs are leveraged to
improve segmentation boundaries between the
categories “Road”, “Sidewalk” and “Wall”.
Geometric Unary Potentials. Left: To
encourage label changes at 3D curbs or
folds after projection into the image
domain. Right: This constraint is
implemented by pixel unary potentials
inside each minimum bounding disc
around each 2D curb or fold segment m.
26. } Exploit 3D info. to automatically generate very accurate object
segmentations given annotated 3D bounding boxes.
} Formulate it as the one of inference in a binary MRF which exploits
appearance models, stereo and/or noisy point clouds, a repository
of 3D CAD models and topological constraints.
} Segment cars with the accuracy of 86% intersection-over-union,
performing as well as highly recommended MTurkers!
Auto generate segmentation ground-
truth (bottom) using weak labels (top).
http://www.cs.utoronto.ca/~fidler/papers/chen_et_al_cvpr14b.pdf
27. (a) CAD model projected to image
plane, (b) contour, (c,d) distance and
signed distance transform.
(Top box) Points falling inside the ground-truth 3D boxes are white, and black outside.
(Bottom box) Point clouds are averaged over the dataset for 8 different viewpoints.
28. } Automatically annotate 3D objects of interest in point clouds of road
scenes by exploiting a multitude of annotated images in image
databases, such as LabelMe and ImageNet.
} An object detector rained on the annotated images is used to locate
the object regions in acquired multi-view images.
} Then, based on the correspondences between multi-view images
and 3D point clouds, a probabilistic graphical model is used to
model the temporal, spatial and geometric constraints to extract the
3D objects automatically.
(a) mobile LiDAR system
(b) multi-view images
https://aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14579/14223
29. To model the temporal, spatial and geometric constraints
by a Markov Random Field (MRF) model.
40. backtracking algorithm
(1) a denoising pointwise segmentation
strategy enabling a fast implementation
of one-click annotation, (2) expand the
motion model technique with guided-
tracking algorithm, easing the frame-to-
frame annotation processes, and (3) an
interactive yet robust open-source point
cloud annotation tool, targeting both
skilled and crowdsourcing annotators to
create high-quality bounding box
annotations.