Matches in SemOpenAlex for { <https://semopenalex.org/work/W3125870003> ?p ?o ?g. }
Showing items 1 to 77 of
77
with 100 items per page.
- W3125870003 abstract "Nowadays, many of the images captured are observed by machines only and not by humans, for example, robots' or autonomous cars' cameras. High-level machine vision models, such as object recognition, assume images are transformed to some canonical image space by the camera ISP. However, the camera ISP is optimized for producing visually pleasing images to human observers and not for machines, thus, one may spare the ISP compute time and apply the vision models directly to the raw data. Yet, it has been shown that training such models directly on the RAW images results in a performance drop. To mitigate this drop in performance (without the need to annotate RAW data), we use a dataset of RAW and RGB image pairs, which can be easily acquired with no human labeling. We then train a model that is applied directly to the RAW data by using knowledge distillation such that the model predictions for RAW images will be aligned with the predictions of an off-the-shelf pre-trained model for processed RGB images. Our experiments show that our performance on RAW images is significantly better than a model trained on labeled RAW images. It also reasonably matches the predictions of a pre-trained model on processed RGB images, while saving the ISP compute overhead." @default.
- W3125870003 created "2021-02-01" @default.
- W3125870003 creator A5025410418 @default.
- W3125870003 creator A5050323834 @default.
- W3125870003 creator A5072571599 @default.
- W3125870003 date "2021-01-25" @default.
- W3125870003 modified "2023-09-27" @default.
- W3125870003 title "ISP Distillation" @default.
- W3125870003 cites W2194775991 @default.
- W3125870003 cites W2294370754 @default.
- W3125870003 cites W2473930607 @default.
- W3125870003 cites W2552290192 @default.
- W3125870003 cites W2615178951 @default.
- W3125870003 cites W2799265886 @default.
- W3125870003 cites W2952046917 @default.
- W3125870003 cites W2963163009 @default.
- W3125870003 cites W2977160928 @default.
- W3125870003 cites W2982247743 @default.
- W3125870003 cites W3012451604 @default.
- W3125870003 cites W3098639991 @default.
- W3125870003 hasPublicationYear "2021" @default.
- W3125870003 type Work @default.
- W3125870003 sameAs 3125870003 @default.
- W3125870003 citedByCount "0" @default.
- W3125870003 crossrefType "posted-content" @default.
- W3125870003 hasAuthorship W3125870003A5025410418 @default.
- W3125870003 hasAuthorship W3125870003A5050323834 @default.
- W3125870003 hasAuthorship W3125870003A5072571599 @default.
- W3125870003 hasConcept C115961682 @default.
- W3125870003 hasConcept C127413603 @default.
- W3125870003 hasConcept C132964779 @default.
- W3125870003 hasConcept C154945302 @default.
- W3125870003 hasConcept C194648553 @default.
- W3125870003 hasConcept C199360897 @default.
- W3125870003 hasConcept C31972630 @default.
- W3125870003 hasConcept C41008148 @default.
- W3125870003 hasConcept C78519656 @default.
- W3125870003 hasConcept C82990744 @default.
- W3125870003 hasConcept C90509273 @default.
- W3125870003 hasConceptScore W3125870003C115961682 @default.
- W3125870003 hasConceptScore W3125870003C127413603 @default.
- W3125870003 hasConceptScore W3125870003C132964779 @default.
- W3125870003 hasConceptScore W3125870003C154945302 @default.
- W3125870003 hasConceptScore W3125870003C194648553 @default.
- W3125870003 hasConceptScore W3125870003C199360897 @default.
- W3125870003 hasConceptScore W3125870003C31972630 @default.
- W3125870003 hasConceptScore W3125870003C41008148 @default.
- W3125870003 hasConceptScore W3125870003C78519656 @default.
- W3125870003 hasConceptScore W3125870003C82990744 @default.
- W3125870003 hasConceptScore W3125870003C90509273 @default.
- W3125870003 hasLocation W31258700031 @default.
- W3125870003 hasOpenAccess W3125870003 @default.
- W3125870003 hasPrimaryLocation W31258700031 @default.
- W3125870003 hasRelatedWork W1975533357 @default.
- W3125870003 hasRelatedWork W2573092587 @default.
- W3125870003 hasRelatedWork W2885160662 @default.
- W3125870003 hasRelatedWork W2888632004 @default.
- W3125870003 hasRelatedWork W2941119526 @default.
- W3125870003 hasRelatedWork W2944756272 @default.
- W3125870003 hasRelatedWork W2950130553 @default.
- W3125870003 hasRelatedWork W2952323569 @default.
- W3125870003 hasRelatedWork W2965187458 @default.
- W3125870003 hasRelatedWork W3017319489 @default.
- W3125870003 hasRelatedWork W3078783273 @default.
- W3125870003 hasRelatedWork W3104713581 @default.
- W3125870003 hasRelatedWork W3112677735 @default.
- W3125870003 hasRelatedWork W3118800911 @default.
- W3125870003 hasRelatedWork W3143049674 @default.
- W3125870003 hasRelatedWork W3144785864 @default.
- W3125870003 hasRelatedWork W3150847003 @default.
- W3125870003 hasRelatedWork W3159201570 @default.
- W3125870003 hasRelatedWork W3160084959 @default.
- W3125870003 hasRelatedWork W3204340825 @default.
- W3125870003 isParatext "false" @default.
- W3125870003 isRetracted "false" @default.
- W3125870003 magId "3125870003" @default.
- W3125870003 workType "article" @default.