BigBIRD: A large-scale 3D database of object instances

By Arjun Singh, James Sha, Karthik S. Narayan, Tudor Achim, Pieter Abbeel. 2014

378 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Geology, 3D Vision & Pose

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

In Geology

In 3D Vision & Pose

Citations in the corpus

Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data (2019, background): “Some datasets [36, 5] are captured in controlled environment which might greatly differ from real-world scenes.”
CORe50: a New Dataset and Benchmark for Continuous Object Recognition (2017, neutral): “ing Outdoor sessions NORB [12] 5 25 20 20* grayscale turntable no COIL-100 [18] - 100 20 54* RGB turntable no iLab-20M [2] 15 704 - - RGB turntable no RGB-D [24] 51 300 - - RGB-D turntable no BigBIRD [25] - 100 - - RGB-D turntable no ALOI [8] - 1000 - - RGB turntable no BigBrother [6] - 7 54 ˘20 RGB”
RGB-D datasets using microsoft kinect or similar sensors: a survey (2016, background): “More details about Big BIRD can be found in [80].”
Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling (2018, background): “Among those, many object datasets are designed for benchmarking robot manipulation [6, 23, 33, 51].”
RLBench: The Robot Learning Benchmark & Learning Environment (2019, background): “Moving on from whole-system benchmarks, there are a host of benchmarks that focus on sub-problems, for example perception datasets, from both the computer vision community (such as ILSVRC [1], COCO [30], Pascal-VOC [31], etc), and the robotics community (such as BigBIRD [32], YCB-Video [33], etc).”
A Large Dataset of Object Scans (2016, methodology): “The BigBIRD dataset [6] provides 3D scans of 100 objects, acquired in the lab using a computer-controlled photobench and a static calibrated imaging setup.”
Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detection (2017, methodology): “Images of objects from different viewpoints: We choose the objects present in Big Berkeley Instance Recognition Dataset (BigBIRD) [45] to conduct our experiments.”
Objaverse: A Universe of Annotated 3D Objects (2022, background): “Datasets such as KIT [33], YCB [5], BigBIRD [64], IKEA [40], and Pix3D [66] provide image-calibrated models over a diverse set of household objects, but severely lack in scale with only a few hundred objects at most.”
iLab-20M: A Large-Scale Controlled Object Dataset to Investigate Deep Learning (2016, methodology): “COIL [41] Handheld 100 1 1 72 Implicit No 7,200 SOIL-47 [29] Handheld — 47 1 42 Implicit No 1,974 Pascal [14] Misc 20 790-10,129 1 1 Yes Partial 11,540 Caltech-101 [15] Google 102 31-800 (μ = 90) 1 1 No No 9,144 Caltech-256 [22] Google 257 80-827 (μ = 119) 1 1 No No 30,607 LabelMe [48] Misc 900 ? ∼1”
DepthSynth: Real-Time Realistic Synthetic Data Generation from CAD Models for 2.5D Recognition (2017, background): “While real data is commonly used for comparison and training, a large number of these recent studies decompose the problems to matching acquired depth images of realworld objects to synthetic ones rendered from a database of pre-existing 3D models [13, 38, 45, 27, 46, 44, 6].”
Range sensor and silhouette fusion for high-quality 3D Scanning (2015, methodology): “The KinectFusion algorithm assigns poses to each incoming camera frame via frameto-model, point-to-plane ICP [22]; because KinectFusion requires slow camera pose movements, and the BigBIRD dataset has 5 cameras in relatively far away locations (see Figure 2c), we use the camera poses that the BigBIR”
F-SIOL-310: A Robotic Dataset and Benchmark for Few-Shot Incremental Object Learning (2021, background): “Further, NORB only has 5 object classes and 25 total objects, while COIL-100 and BigBIRD do not provide object classes and contain only 100 objects in total, which is extremely small compared to other datasets like CIFAR-100 or OpenLORIS.”
RGBD Datasets: Past, Present and Future (2016, methodology): “= camera pose computed from RGBD data; = camera pose from calibration system. c Captured using the same turntable setup as the BigBIRD dataset.”
Video Object Segmentation using Teacher-Student Adaptation in a Human Robot Interaction (HRI) Setting (2018, neutral): “RGB-D [25] - 51 300 T BIG BIRD [26] - - 100 T ICUB 28 [20] 4 7 28 H ICUB World [21] 6 20 200 H Core50 [17] 11 10 50 H IVOS 12 12 36 H”
A survey on deep multimodal learning for computer vision: advances, trends, applications, and datasets (2021, background): “– BigBIRD:The dataset was originally introduced by [56].”
ABO: Dataset and Benchmarks for Real-World 3D Object Understanding (2021, background): “Multi-View Cars [24] and BigBIRD [49] are real-world object-focused multi-view datasets but have limited numbers of instances and categories.”
Benchmarking in Manipulation Research: The YCB Object and Model Set and Benchmarking Protocols (2015, methodology): “[1-4] with a thorough overview provided in section II), generally for object recognition or planning purposes.”
A self-supervised learning system for object detection using physics simulation and multi-view pose estimation (2017, methodology): “The recent Amazon Picking Challenge (APC) [2] has reinforced this realization and has led into the development of datasets specifically for the detection of objects inside shelving units [3], [4], [5].”
Grasp Pose Detection in Point Clouds (2017, methodology): “X) in the reference frame of the hand. Our representation is illustrated in Figure 5. Figure 5(a) shows a grasp candidate generated with respect to partial point cloud data (from the BigBird dataset (Singh et al. 2014)). Figure 5(b) shows two sets of points in R. One set of points, shown in magenta,”