Learning to localize detected objects

By Qieyun Dai, Derek Hoiem. 2012

48 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

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

In Computer Vision and Pattern Recognition

Citations in the corpus

Microsoft COCO: Common Objects in Context (2014, methodology): “Generating segmentations from detections We now describe a simple method for generating object bounding boxes and segmentation masks, following prior work that produces segmentations from object detections [47], [48], [49], [50].”
Hypercolumns for object segmentation and fine-grained localization (2014, background): “Dai and Hoiem [12] generalize this reasoning to all categories.”
Simultaneous Detection and Segmentation (2014, background): “Dai and Hoiem generalize this to all categories and use instance and category appearance models to improve detection [7].”
Watch and learn: Semi-supervised learning of object detectors from videos (2015, methodology): “We believe methods like [10] can be used to further improve the labeling quality.”
segDeepM: Exploiting segmentation and context in deep neural networks for object detection (2015, methodology): “[5] uses DPM to find a rough object location and refines it according to color information and occlusion boundaries.”
Improving object detection with deep convolutional networks via Bayesian optimization and structured prediction (2015, methodology): “Dai and Hoiem [7] used the structured SVM with color and edge features to refine the bounding box coordinates in DPM framework.”
Beyond Bounding Boxes: Precise Localization of Objects in Images (2015, background): “Dai and Hoiem [14] generalize this reasoning to all categories.”
Deep learning for visual understanding: A review (2016, background): “Although those schemes are good at ﬁ nding rough object positions, they normally could not accurately localize the whole object via a tight bounding box, which forms the largest source of detection error [180,181].”
Efficient Object Detection and Segmentation for Fine-Grained Recognition (2013, background): “For example, images are taken in natural settings with rich and challenging backgrounds, where the background features may become prominent and serve as distractors to the recognition algorithm.”
Towards Unified Object Detection and Semantic Segmentation (2014, unknown): “utilized segments extracted for each object detection hypothesis for better localization [12].”
Object Localization based on Structural SVM using Privileged Information (2014, unknown): “More recently, [19, 20] employ SSVM as part of their localization procedure, however none of them incorporate privileged information or similar idea.”
Bottom-Up Segmentation for Top-Down Detection (2013, background): “Dai et al. [11] exploit a DPM to find a rough location for the object of interest and refine the detected bounding box according to occlusion boundaries and color information.”
SCALPEL: Segmentation Cascades with Localized Priors and Efficient Learning (2013, methodology): “Most importantly, where [14, 10] simply transfer a matched object mask and [6, 15] learn to segment on pixels using graph-cut and sub-modular edge features, we learn to greedily segment on superpixels with arbitrary features, a cascaded weight vector, and with a set of class- and scale-specific mode”
Deep learning for visual understanding (2017, unknown): “Although those schemes are good at nding rough object positions, they normally could not accurately localize the whole object via a tight bounding box, which forms the largest source of detection error [163, 164].”
Amodal Instance Segmentation (2016, methodology): “Earlier approaches rely on DPM [11] detections and predict segmentation masks using a simple appearance model [36,30,8], combine DPM detections and semantic segmentation predictions [12,9] or adopt a transductive approach [34].”