Detection and Retrieval of Out-of-Distribution Objects in Semantic Segmentation

Detection and Retrieval of Out-of-Distribution Objects in Semantic Segmentation. Oberdiek, P., Rottmann, M., & Fink, G., A. 2020.

Paper abstract bibtex

When deploying deep learning technology in self-driving cars, deep neural networks are constantly exposed to do- main shifts. These include, e.g., changes in weather con- ditions, time of day, and long-term temporal shift. In this work we utilize a deep neural network trained on the Cityscapes dataset containing urban street scenes and in- fer images from a different dataset, the A2D2 dataset, con- taining also countryside and highway images. We present a novel pipeline for semantic segmenation that detects out- of-distribution (OOD) segments by means of the deep neu- ral network’s prediction and performs image retrieval af- ter feature extraction and dimensionality reduction on im- age patches. In our experiments we demonstrate that the deployed OOD approach is suitable for detecting out-of- distribution concepts. Furthermore, we evaluate the im- age patch retrieval qualitatively as well as quantitatively by means of the semi-compatible A2D2 ground truth and obtain mAP values ofup to 52.2%. 1.

@misc{
 title = {Detection and Retrieval of Out-of-Distribution Objects in Semantic Segmentation},
 type = {misc},
 year = {2020},
 pages = {328-329},
 id = {e476a771-167d-3186-b356-b9ff85b377da},
 created = {2022-10-31T09:30:48.681Z},
 accessed = {2022-10-31},
 file_attached = {true},
 profile_id = {235249c2-3ed4-314a-b309-b1ea0330f5d9},
 group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},
 last_modified = {2023-04-03T10:17:19.258Z},
 read = {true},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 folder_uuids = {5a010301-acb6-4642-a6b2-8afaee1b741c,4b11b59b-846d-45b5-82b7-81d8bc3108b2,bd3c6f2e-3514-47cf-bc42-12db8b9abe45},
 private_publication = {false},
 abstract = {When deploying deep learning technology in self-driving cars, deep neural networks are constantly exposed to do- main shifts. These include, e.g., changes in weather con- ditions, time of day, and long-term temporal shift. In this work we utilize a deep neural network trained on the Cityscapes dataset containing urban street scenes and in- fer images from a different dataset, the A2D2 dataset, con- taining also countryside and highway images. We present a novel pipeline for semantic segmenation that detects out- of-distribution (OOD) segments by means of the deep neu- ral network’s prediction and performs image retrieval af- ter feature extraction and dimensionality reduction on im- age patches. In our experiments we demonstrate that the deployed OOD approach is suitable for detecting out-of- distribution concepts. Furthermore, we evaluate the im- age patch retrieval qualitatively as well as quantitatively by means of the semi-compatible A2D2 ground truth and obtain mAP values ofup to 52.2%. 1.},
 bibtype = {misc},
 author = {Oberdiek, Philipp and Rottmann, Matthias and Fink, Gernot A.}
}

Downloads: 0

{"_id":"5A7YxYpdfdv6dCL53","bibbaseid":"oberdiek-rottmann-fink-detectionandretrievalofoutofdistributionobjectsinsemanticsegmentation-2020","author_short":["Oberdiek, P.","Rottmann, M.","Fink, G., A."],"bibdata":{"title":"Detection and Retrieval of Out-of-Distribution Objects in Semantic Segmentation","type":"misc","year":"2020","pages":"328-329","id":"e476a771-167d-3186-b356-b9ff85b377da","created":"2022-10-31T09:30:48.681Z","accessed":"2022-10-31","file_attached":"true","profile_id":"235249c2-3ed4-314a-b309-b1ea0330f5d9","group_id":"5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1","last_modified":"2023-04-03T10:17:19.258Z","read":"true","starred":false,"authored":false,"confirmed":false,"hidden":false,"folder_uuids":"5a010301-acb6-4642-a6b2-8afaee1b741c,4b11b59b-846d-45b5-82b7-81d8bc3108b2,bd3c6f2e-3514-47cf-bc42-12db8b9abe45","private_publication":false,"abstract":"When deploying deep learning technology in self-driving cars, deep neural networks are constantly exposed to do- main shifts. These include, e.g., changes in weather con- ditions, time of day, and long-term temporal shift. In this work we utilize a deep neural network trained on the Cityscapes dataset containing urban street scenes and in- fer images from a different dataset, the A2D2 dataset, con- taining also countryside and highway images. We present a novel pipeline for semantic segmenation that detects out- of-distribution (OOD) segments by means of the deep neu- ral network’s prediction and performs image retrieval af- ter feature extraction and dimensionality reduction on im- age patches. In our experiments we demonstrate that the deployed OOD approach is suitable for detecting out-of- distribution concepts. Furthermore, we evaluate the im- age patch retrieval qualitatively as well as quantitatively by means of the semi-compatible A2D2 ground truth and obtain mAP values ofup to 52.2%. 1.","bibtype":"misc","author":"Oberdiek, Philipp and Rottmann, Matthias and Fink, Gernot A.","bibtex":"@misc{\n title = {Detection and Retrieval of Out-of-Distribution Objects in Semantic Segmentation},\n type = {misc},\n year = {2020},\n pages = {328-329},\n id = {e476a771-167d-3186-b356-b9ff85b377da},\n created = {2022-10-31T09:30:48.681Z},\n accessed = {2022-10-31},\n file_attached = {true},\n profile_id = {235249c2-3ed4-314a-b309-b1ea0330f5d9},\n group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},\n last_modified = {2023-04-03T10:17:19.258Z},\n read = {true},\n starred = {false},\n authored = {false},\n confirmed = {false},\n hidden = {false},\n folder_uuids = {5a010301-acb6-4642-a6b2-8afaee1b741c,4b11b59b-846d-45b5-82b7-81d8bc3108b2,bd3c6f2e-3514-47cf-bc42-12db8b9abe45},\n private_publication = {false},\n abstract = {When deploying deep learning technology in self-driving cars, deep neural networks are constantly exposed to do- main shifts. These include, e.g., changes in weather con- ditions, time of day, and long-term temporal shift. In this work we utilize a deep neural network trained on the Cityscapes dataset containing urban street scenes and in- fer images from a different dataset, the A2D2 dataset, con- taining also countryside and highway images. We present a novel pipeline for semantic segmenation that detects out- of-distribution (OOD) segments by means of the deep neu- ral network’s prediction and performs image retrieval af- ter feature extraction and dimensionality reduction on im- age patches. In our experiments we demonstrate that the deployed OOD approach is suitable for detecting out-of- distribution concepts. Furthermore, we evaluate the im- age patch retrieval qualitatively as well as quantitatively by means of the semi-compatible A2D2 ground truth and obtain mAP values ofup to 52.2%. 1.},\n bibtype = {misc},\n author = {Oberdiek, Philipp and Rottmann, Matthias and Fink, Gernot A.}\n}","author_short":["Oberdiek, P.","Rottmann, M.","Fink, G., A."],"urls":{"Paper":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c/file/7ab80b46-51f7-79d9-0d8e-59841a71fed2/full_text.pdf.pdf"},"biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","bibbaseid":"oberdiek-rottmann-fink-detectionandretrievalofoutofdistributionobjectsinsemanticsegmentation-2020","role":"author","metadata":{"authorlinks":{}},"downloads":0},"bibtype":"misc","biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","dataSources":["2252seNhipfTmjEBQ"],"keywords":[],"search_terms":["detection","retrieval","out","distribution","objects","semantic","segmentation","oberdiek","rottmann","fink"],"title":"Detection and Retrieval of Out-of-Distribution Objects in Semantic Segmentation","year":2020}