CoChat
Sign Up
Cited by 65Advanced Neural Network ApplicationsVisual Attention and Saliency DetectionAdvanced Image and Video Retrieval Techniques

The DOI matches its Crossref record; 3 of the 3 references it cites were found in a registry or an index.

Object-Proposal Evaluation Protocol is ‘Gameable’

Open access
Chavali, Neelima, Agrawal, Harsh, Mahendru, Aroma, Batra, Dhruv·2016·2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
DOI 10.1109/cvpr.2016.97
Open this paper in CoChat — free accountOpen PDF

Abstract

Object proposals have quickly become the de-facto preprocessing step in a number of vision pipelines (for object detection, object discovery, and other tasks). Their performance is usually evaluated on partially annotated datasets. In this paper, we argue that the choice of using a partially annotated dataset for evaluation of object proposals is problematic - as we demonstrate via a thought experiment, the evaluation protocol is 'gameable', in the sense that progress under this protocol does not necessarily correspond to a "better" category independent object proposal algorithm. To alleviate this problem, we: (1) Introduce a nearly-fully annotated version of PASCAL VOC dataset, which serves as a test-bed to check if object proposal techniques are overfitting to a particular list of categories. (2) Perform an exhaustive evaluation of object proposal methods on our introduced nearly-fully annotated PASCAL dataset and perform cross-dataset generalization experiments, and (3) Introduce a diagnostic experiment to detect the bias capacity in an object proposal algorithm. This tool circumvents the need to collect a densely annotated dataset, which can be expensive and cumbersome to collect. Finally, we have released an easy-to-use toolbox which combines various publicly available implementations of object proposal algorithms which standardizes the proposal generation and evaluation so that new methods can be added and evaluated on different datasets. We hope that the results presented in the paper will motivate the community to test the category independence of various object proposal methods by carefully choosing the evaluation protocol.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1109/cvpr.2016.97

Cites

3 of 3 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The Recall-to-IoU metric is just loosely related to the ultimate detection accuracy.

    “It is noteworthy that the Recall-to-IoU metric is just loosely [19], [20], [21] related to the ultimate detection accuracy.”

    From Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (2017) · cites this paper as [21]

    Section 3, “Evaluating Object Proposals” (page unavailable), says average recall correlates with ultimate detection performance better than other metrics. The paper also evaluates Recall-vs-IoU in section 8.3, but it does not establish that this particular metric is only loosely related to detection accuracy. (LLM, conf=0.78)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 56

  1. sun, 2011, Scale and object aware image retargeting for thumbnail browsing, ICCV
  2. 10.1109/TPAMI.2011.158 · DOI
  3. 10.1007/978-3-642-33783-3_4 · DOI
  4. 10.1109/CVPR.2013.253 · DOI
  5. 10.1109/ICCV.2015.363 · DOI
  6. wang, 2015, Transferring Rich Feature Hierarchies for Robust Visual Tracking
  7. 10.1007/978-3-642-33885-4_27 · DOI
  8. 10.1109/ICCV.2015.181 · DOI
  9. 10.1109/CVPR.2014.119 · DOI · Paper page
  10. 10.1109/TPAMI.2009.167 · DOI · Paper page
  11. cho, 2015, Unsupervised object discovery and localization in the wild Part-based matching with bottom-up region proposals CoRR
  12. lin, 2014, Microsoft COCO: Common objects in context, ECCV
  13. 10.1007/s11263-012-0538-3 · DOI
  14. 10.1109/CVPR.2015.7299063 · DOI
  15. 10.1109/TPAMI.2013.122 · DOI
  16. zitnick, 2014, Edge boxes: Locating object proposals from edges, ECCV
  17. zhu, 2015, segdeepm: Exploiting segmentation and context in deep neural networks for object detection, CVPR
  18. 10.1109/TPAMI.2011.231 · DOI
  19. 10.1109/ICCV.2015.40 · DOI
  20. carreira, 2012, Semantic segmentation with second-order pooling, ECCV
  21. 10.1109/CVPR.2012.6248077 · DOI
  22. 10.1109/CVPR.2015.7298987 · DOI
  23. 10.1109/CVPR.2015.7299025 · DOI
  24. oneata, 2014, Spatio-temporal object detection proposals, ECCV 2014
  25. 10.1109/CVPR.2015.7298974 · DOI
  26. banica, 2013, CPMC-3D-02P semantic segmentation of RGB-D images using CPMC and second order pooling
  27. gupta, 2014, Learning Rich Features from RGB-D Images for Object Detection and Segmentation, 10.1007/978-3-319-10584-0_23 · DOI · Paper page
  28. 10.1109/CVPR.2015.7299047 · DOI
  29. 10.1109/CVPR.2015.7298982 · DOI
  30. 10.1109/CVPR.2015.7298735 · DOI
  31. 10.1109/CVPR.2014.414 · DOI
  32. krähenbühl, 2014, Geodesic object proposals, ECCV
  33. russakovsky, 2014, Imagenet Large Scale Visual Recognition Challenge
  34. girshick, 2013, Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
  35. he, 2014, Spatial pyramid pooling in deep convolutional networks for visual recognition
  36. szegedy, 2014, Scalable high-quality object detection
  37. 10.1109/ICCV.2013.10 · DOI
  38. 10.1109/ICCV.2013.369 · DOI
  39. erhan, 2013, Scalable object detection using deep neural networks In CVPR
  40. 10.1109/CVPR.2015.7298597 · DOI
  41. 10.1109/CVPR.2015.7298673 · DOI
  42. 10.1109/TPAMI.2012.28 · DOI
  43. 10.1109/CVPR.2014.49 · DOI · Paper page
  44. 10.1109/ICCV.2013.315 · DOI
  45. 10.1109/ICCV.2011.6126351 · DOI
  46. 10.1007/s11263-013-0620-5 · DOI · Paper page
  47. 10.1109/CVPR.2014.310 · DOI
  48. 10.1109/CVPR.2015.7299012 · DOI
  49. humayun, 2014, Rigor- recycling inference in graph cuts for generating object regions, CVPR
  50. pinheiro, 2015, Learning to segment object candidates
  51. 10.1109/ICCV.2015.296 · DOI
  52. chen, 2015, Improving object proposals with multi-thresholding straddling expansion, CVPR
  53. krähenbühl, 2015, Learning to propose objects, CVPR
  54. heitz, 2008, Learning spatial context: Using stuff to find things, Proc 10th European Conference on Computer Vision
  55. kuo, 2015, Deepbox Learning objectness with convolutional networks
  56. 10.1109/TPAMI.2014.2315811 · DOI

Cited by

Sorted by citations

Metadata

Authors
Chavali, N., Agrawal, H., Mahendru, A., & Batra, D.
Affiliations
  • Amgen (United States) (US)
  • Virginia Tech (US)
ORCID
Venue
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
Publisher
IEEE
Published
2016-06
Pages
835-844
ISSN
1063-6919 (ISSN-L)
DOI
10.1109/cvpr.2016.97
OpenAlex
W1790312427
Cited by
65
References
3
Citations by year
2025: 3 · 2024: 1 · 2023: 4 · 2022: 5 · 2021: 5 · 2020: 4 · 2019: 9 · 2018: 13 · 2017: 10 · 2016: 6
FWCI
4.40
Type
conference paper
Language
en
Open access
Locations
  • IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Visual Attention and Saliency Detection
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
Keywords
object proposals, evaluation protocol, overfitting, Pascal VOC, cross-dataset generalization, bias detection, evaluation benchmarking

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review