On the Importance of Label Quality for Semantic Segmentation
Convolutional networks (ConvNets) have become the dominant approach to semantic image segmentation. Producing accurate, pixel-level labels required for this task is a tedious and time consuming process; however, producing approximate, coarse labels could take only a fraction of the time and effort....
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020
|
| Online Access: | https://hdl.handle.net/1721.1/124403 |
| _version_ | 1826216795190591488 |
|---|---|
| author | Zlateski, Aleksandar Jaroensri, Ronnachai Sharma, Prafull Durand, Frederic |
| author2 | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science |
| author_facet | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Zlateski, Aleksandar Jaroensri, Ronnachai Sharma, Prafull Durand, Frederic |
| author_sort | Zlateski, Aleksandar |
| collection | MIT |
| description | Convolutional networks (ConvNets) have become the dominant approach to semantic image segmentation. Producing accurate, pixel-level labels required for this task is a tedious and time consuming process; however, producing approximate, coarse labels could take only a fraction of the time and effort. We investigate the relationship between the quality of labels and the performance of ConvNets for semantic segmentation. We create a very large synthetic dataset with perfectly labeled street view scenes. From these perfect labels, we synthetically coarsen labels with different qualities and estimate human-hours required for producing them. We perform a series of experiments by training ConvNets with a varying number of training images and label quality. We found that the performance of ConvNets mostly depends on the time spent creating the training labels. That is, a larger coarsely-annotated dataset can yield the same performance as a smaller finely-annotated one. Furthermore, fine-tuning coarsely pre-trained ConvNets with few finely-annotated labels can yield comparable or superior performance to training it with a large amount of finely-annotated labels alone, at a fraction of the labeling cost. We demonstrate that our result is also valid for different network architectures, and various object classes in an urban scene. |
| first_indexed | 2024-09-23T16:53:14Z |
| format | Article |
| id | mit-1721.1/124403 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T16:53:14Z |
| publishDate | 2020 |
| publisher | IEEE |
| record_format | dspace |
| spelling | mit-1721.1/1244032022-10-03T08:58:07Z On the Importance of Label Quality for Semantic Segmentation Zlateski, Aleksandar Jaroensri, Ronnachai Sharma, Prafull Durand, Frederic Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Convolutional networks (ConvNets) have become the dominant approach to semantic image segmentation. Producing accurate, pixel-level labels required for this task is a tedious and time consuming process; however, producing approximate, coarse labels could take only a fraction of the time and effort. We investigate the relationship between the quality of labels and the performance of ConvNets for semantic segmentation. We create a very large synthetic dataset with perfectly labeled street view scenes. From these perfect labels, we synthetically coarsen labels with different qualities and estimate human-hours required for producing them. We perform a series of experiments by training ConvNets with a varying number of training images and label quality. We found that the performance of ConvNets mostly depends on the time spent creating the training labels. That is, a larger coarsely-annotated dataset can yield the same performance as a smaller finely-annotated one. Furthermore, fine-tuning coarsely pre-trained ConvNets with few finely-annotated labels can yield comparable or superior performance to training it with a large amount of finely-annotated labels alone, at a fraction of the labeling cost. We demonstrate that our result is also valid for different network architectures, and various object classes in an urban scene. 2020-03-30T13:45:05Z 2020-03-30T13:45:05Z 2018-12 2019-05-29T13:33:27Z Article http://purl.org/eprint/type/ConferencePaper 9781538664209 https://hdl.handle.net/1721.1/124403 Zlateski, Aleksandar et al. "On the Importance of Label Quality for Semantic Segmentation." 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 18-23, 2018, Salt Lake City, UT, IEEE, December 2018. © 2018 IEEE. en http://dx.doi.org/10.1109/cvpr.2018.00160 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf IEEE Computer Vision Foundation |
| spellingShingle | Zlateski, Aleksandar Jaroensri, Ronnachai Sharma, Prafull Durand, Frederic On the Importance of Label Quality for Semantic Segmentation |
| title | On the Importance of Label Quality for Semantic Segmentation |
| title_full | On the Importance of Label Quality for Semantic Segmentation |
| title_fullStr | On the Importance of Label Quality for Semantic Segmentation |
| title_full_unstemmed | On the Importance of Label Quality for Semantic Segmentation |
| title_short | On the Importance of Label Quality for Semantic Segmentation |
| title_sort | on the importance of label quality for semantic segmentation |
| url | https://hdl.handle.net/1721.1/124403 |
| work_keys_str_mv | AT zlateskialeksandar ontheimportanceoflabelqualityforsemanticsegmentation AT jaroensrironnachai ontheimportanceoflabelqualityforsemanticsegmentation AT sharmaprafull ontheimportanceoflabelqualityforsemanticsegmentation AT durandfrederic ontheimportanceoflabelqualityforsemanticsegmentation |