Submodular Object Recognition

Abstract

We present a novel object recognition framework based on multiple figure-ground hypotheses with a large object spatial support, generated by bottom-up processes and midlevel cues in an unsupervised manner. We exploit the benefit of regression for discriminating segments' categories and qualities, where a regressor is trained to each category using the overlapping observations between each figureground segment hypothesis and the ground-truth of the target category in an image. Object recognition is achieved by maximizing a submodular objective function, which maximizes the similarities between the selected segments (i.e., facility locations) and their group elements (i.e., clients), penalizes the number of selected segments, and more importantly, encourages the consistency of object categories corresponding to maximum regression values from different category-specific regressors for the selected segments. The proposed framework achieves impressive recognition results on three benchmark datasets, including PASCAL VOC 2007, Caltech-101 and ETHZ-shape.

Image Location Estimation by Salient Region Matching

“…Our image location estimation is purely based on image content. We can convert the image location estimation problem as content based image retrieval [25], [55]- [57], [64] or object recognition problem [58]- [61]. In social media community, the shared photos are always attached with tags, time stamps, user's comments, and geo-coordinates that images are taken.…”

An incremental structured part model for object recognition

“…We first make this comparison against five other classification methods [34,33,36,37,19] Table 1 shows the classification accuracy on all 20 classes. The latest method proposed in [19] has achieved the highest accuracy in most classes.…”

“…The latest method proposed in [19] has achieved the highest accuracy in most classes. However, our method still achieve considerable results with other state-of-the-art classification methods, especially those with similar shapes such as bicycle and motorbike, cat and dog, cow and sheep.…”

“…One noteworthy observation is that a most recently proposed method, namely submodular object recognition [19], has achieved the best classification performance on benchmark evaluation. Unlike the submodular model which conducts object recognition at pixel level, our framework emphasizes the exploration of object part relationships and the ability of model updates.…”

Multi-task proximal support vector machine

“…Various image classification and recognition methods have been proposed and have achieved much success [1,2,3,4,5,6,7,8,9]. Some feature learning methods are also proposed to improve the performance of image classification and recognition [10,11,12,13].…”