What leads to generalization of object proposals?

By Rui Wang, Dhruv Kumar Mahajan, Vignesh Ramanathan. 2020

12 citations recorded by the source corpus.

Object proposal generation is often the first step in many detection models. It is lucrative to train a good proposal model, that generalizes to unseen classes. This could help scaling detection models to larger number of classes with fewer annotations. Motivated by this, we study how a detection model trained on a small set of source classes can provide proposals that generalize to unseen classes. We systematically study the properties of the dataset - visual diversity and label space granularity - required for good generalization. We show the trade-off between using fine-grained labels and coarse labels. We introduce the idea of prototypical classes: a set of sufficient and necessary classes required to train a detection model to obtain generalized proposals in a more data-efficient way. On the Open Images V4 dataset, we show that only 25% of the classes can be selected to form such a prototypical set. The resulting proposals from a model trained with these classes is only 4.3% worse than using all the classes, in terms of average recall (AR). We also demonstrate that Faster R-CNN model leads to better generalization of proposals compared to a single-stage network like RetinaNet.

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

Grounded Language-Image Pre-training (2021, background): “A line of work identifies building a openworld object proposal module that could propose any novel object at test time as the key challenge [22, 25, 60, 74, 75]; GLIP offers a new perspective: the model does not need to propose every possible novel objects from an open set; rather it only needs to p”
Open-World Instance Segmentation: Exploiting Pseudo Ground Truth From Learned Pairwise Affinity (2022, neutral): “Open-world setup [5, 43] has been recently (re)-introduced into various problems in computer vision such as recognition [30, 38], tracking [36], detection [25, 27, 53], and segmentation [54].”
Learning Open-World Object Proposals without Learning to Classify (2021, methodology): “[64] study the generalization from a dataset standpoint and demonstrate the impact of visual diversity and label granularity of training dataset on the generalization of object proposers.”
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.”