DeepBox: Learning Objectness with Convolutional Networks

By Weicheng Kuo, Bharath Hariharan, Jitendra Malik. 2015

184 citations recorded by the source corpus.

Existing object proposal approaches use primarily bottom-up cues to rank proposals, while we believe that "objectness" is in fact a high level construct. We argue for a data-driven, semantic approach for ranking object proposals. Our framework, which we call DeepBox, uses convolutional neural networks (CNNs) to rerank proposals from a bottom-up method. We use a novel four-layer CNN architecture that is as good as much larger networks on the task of evaluating objectness while being much faster. We show that DeepBox significantly improves over the bottom-up ranking, achieving the same recall with 500 proposals as achieved by bottom-up methods with 2000. This improvement generalizes to categories the CNN has never seen before and leads to a 4.5-point gain in detection mAP. Our implementation achieves this performance while running at 260 ms per image.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning to Segment Object Candidates (2015, methodology): “Generalization: To see if our approach can generalize to unseen classes [2, 19], we train two additional versions of our model, DeepMask20 and DeepMask20∗.”
What Makes for Effective Detection Proposals? (2015, background): “Indeed, there has been some very recent work in this space [60], [61], [62] that shows promising”
Learning Open-World Object Proposals without Learning to Classify (2021, methodology): “Our main insight is that the binary classification learning in existing object proposers [1], [2], [5] or class-agnostic detectors [13], [14] hinders such generalization, because the model tends to overfit to labeled objects and treat the unlabeled objects in the training set as background.”
Object-Proposal Evaluation Protocol is ‘Gameable’ (2015, background): “These methods include [2, 3, 6–9, 11, 44, 46, 49].”
HyperNet: Towards Accurate Region Proposal Generation and Joint Object Detection (2016, methodology): “Object proposals [5][22][36][4] considerably reduce the computation compared with sliding window methods [10][26] in detection framework.”
Attend Refine Repeat: Active Box Proposal Generation via In-Out Localization (2016, methodology): “ segmentations. Several approaches have been proposed in the literature for this task [1,2,5,6,8,18,25,26,29,40,44]. Among them our work is most related to the CNN-based objectness scoring approaches [12,27,33] that recently have demonstrated state-of-the-art results [32,33]. In the objectness scori”
CRAFT Objects from Images (2016, methodology): “Results show that “DeepBox” is better than “SS”.”
Self-taught object localization with deep networks (2014, methodology): “Recently, convolutional networks were used for region proposal [18, 24].”
Object Detection With Deep Learning: A Review (2018, background): “Also some researches tried to re-rank or refine pre-extracted region proposals to remove unnecessary ones and obtained a limited number of valuable ones, such as DeepBox [83] and SharpMask [84].”
Face Detection with the Faster R-CNN (2016, methodology): “The RPN performs slightly better than DeepBox, perhaps since it uses a deeper CNN. Due to the sharing of convolutional layers between the RPN and the Fast R-CNN detector, the process time of the entire system is lower.”
F-VLM: Open-Vocabulary Object Detection upon Frozen Vision and Language Models (2022, methodology): “Inspired by the learned objectness (Kim et al., 2022; Kuo et al., 2015; Wang et al., 2020), we use class-agnostic box regression and mask prediction heads instead.”
Advanced Deep-Learning Techniques for Salient and Category-Specific Object Detection: A Survey (2018, methodology): “[2] pursued a data-driven, semantic approach for ranking object proposals.”
Unsupervised Learning of Discriminative Attributes and Visual Representations (2016, methodology): “, using re-ranking in [8]) is highly promising for further improvements.”
Review of Deep Learning Algorithms and Architectures (2019, background): “Reference [68] proposed a four-layered CNN termed DeepBox that outperformed larger networks in speed and accuracy.”
Object Detection in 20 Years: A Survey (2019, unknown): “and 3) neural network based approaches [123–128].”
Learning About Objects by Learning to Interact with Them (2020, unknown): “, [66, 33]) and weakly supervised (e.”
Temporal Action Localization in Untrimmed Videos via Multi-stage CNNs (2016, methodology): “All required by Segment-CNN is three deep network models, which occupy less than 1 GB in total.”
Evolution of Image Segmentation using Deep Convolutional Neural Network: A Survey (2020, methodology): “object detection. First stage detects object proposals using Selective Search [63] algorithm and second stage classify those proposals using dierent CNN based classier. Multibox [114, 115], Deepbox [116], Edgebox [117] have used CNN based proposal generation method for object detection. Faster R-CNN”