Citation Context

... detection has made great strides in recent years. However, most state-of-the-art detectors still require each category specific classifiers to evaluate many image windows in a sliding window fashion =-=[17, 25]-=-. In order to reduce the number of windows each classifier needs to consider, training an objectness measure which is generic over categories has recently becomes popular [2,3,21,22,48,49, 57]. Object...

Citation Context

...automatically through the bits which do not belong to bx,y , and makes room to insert rx,y . Our efficient BING feature calculation shares the cumulative nature with the integral image representation =-=[52]-=-. Instead of calculating a single scalar value over an arbitrary rectangle range [52], our method uses a few atomic operations (e.g.ADD, BITWISE, etc.) to calculate a set of binary patterns over an 8×...

Citation Context

...tness proposal generation. Fixation prediction models aim at predicting saliency points of human eye movement [4, 37]. Inspired by neurobiology research about early primate visual system, Itti et al. =-=[36]-=- proposed one of the first computational models for saliency detection, which estimates center-surrounded difference across multi-scale image features. Ma and Zhang [42] proposed a fuzzy growing model...

Citation Context

...The 3 methods [3, 48, 57] have been evaluated on the same benchmark and shown to outperform other alternative proposal methods [6,21,25,30,50], saliency measures [33, 36], interesting point detectors =-=[44]-=-, and HOG detector [17] (see [3] for the comparisons). Best viewed in color. are not used for training. Specifically, we train our method using 6 object categories (i.e. the first 6 categories in VOC ...

Citation Context

... detection has made great strides in recent years. However, most state-of-the-art detectors still require each category specific classifiers to evaluate many image windows in a sliding window fashion =-=[17, 25]-=-. In order to reduce the number of windows each classifier needs to consider, training an objectness measure which is generic over categories has recently becomes popular [2,3,21,22,48,49, 57]. Object...

Citation Context

...ctness measurement with NG To learn an objectness measure of image windows, we follow the general idea of the two stages cascaded SVM [57]. Stage I. We learn a single model w for (1) using linear SVM =-=[24]-=-. NG features of the ground truth object windows and random sampled background windows are used as positive and negative training samples respectively. Stage II. To learn vi and ti in (3) using a line...

Citation Context

...duce the computation for each of them. To the best of our knowledge, no prior method can satisfy all these ambitious goals simultaneously. Research from cognitive psychology [47, 54] and neurobiology =-=[20,38]-=- suggest that humans have a strong ability to perceive objects before identifying them. Based on the human reaction time that is observed and the biological signal transmission time that is estimated,...

Citation Context

...duce the computation for each of them. To the best of our knowledge, no prior method can satisfy all these ambitious goals simultaneously. Research from cognitive psychology [47, 54] and neurobiology =-=[20,38]-=- suggest that humans have a strong ability to perceive objects before identifying them. Based on the human reaction time that is observed and the biological signal transmission time that is estimated,...

Citation Context

...ke, wires, etc. Future works. The high quality and efficiency of our method make it suitable for realtime multi-category object detection applications and large scale image collections (e.g. ImageNet =-=[19]-=-). The binary operations and memory efficiency make our method suitable to run on low power devices [28, 59]. Our speed-up strategy by reducing the number of windows is complementary to other speed-up...

Citation Context

...ed features to obtain greater discrimination, while using advanced speed up techniques to make the computational time tractable. We have extensively evaluated our method on the PASCAL VOC2007 dataset =-=[23]-=-. The experimental results show that our method efficiently (300fps on a single laptop CPU) generates a small set of data-driven, category-independent, high quality object windows, yielding 96.2% dete...

Citation Context

..., and 99.5% DR using 5,000 proposals. The 3 methods [3, 48, 57] have been evaluated on the same benchmark and shown to outperform other alternative proposal methods [6,21,25,30,50], saliency measures =-=[33, 36]-=-, interesting point detectors [44], and HOG detector [17] (see [3] for the comparisons). Best viewed in color. are not used for training. Specifically, we train our method using 6 object categories (i...

Citation Context

...(ε) βj = 〈aj , ε〉/‖aj‖2 (project ε onto aj) ε← ε− βjaj (update residual) end for ilar to the multi-size center-surrounded patterns [36] hypothesized as biologically plausible architecture of primates =-=[27,38,54]-=-. The large weights along the borders of w favor a boundary that separate an object (center) from its background (surrounded). Compared to manually designed center surround patterns [36], our learned ...

Citation Context

...ency detection, which estimates center-surrounded difference across multi-scale image features. Ma and Zhang [42] proposed a fuzzy growing model to analyze local contrast based saliency. Harel et al. =-=[29]-=- proposed normalizing center-surrounded feature maps for highlighting conspicuous parts. Although fixation point prediction models have achieved remarkable development, the prediction results tends to...

Citation Context

...eves 96.2% DR using 1,000 proposals, and 99.5% DR using 5,000 proposals. The 3 methods [3, 48, 57] have been evaluated on the same benchmark and shown to outperform other alternative proposal methods =-=[6,21,25,30,50]-=-, saliency measures [33, 36], interesting point detectors [44], and HOG detector [17] (see [3] for the comparisons). Best viewed in color. are not used for training. Specifically, we train our method ...

Citation Context

...d is 1,000+ times faster than most popular alternatives [3, 22, 48] (see Sec. 4). In addition, for efficient sliding window object detection, keeping the computational cost feasible is very important =-=[43, 51]-=-. Lampert et al. [39] presented an elegant branch-and-bound scheme for detection. However, it can only be used to speed up classifiers that users can provide a good bound on highest score. Also, some ...

Citation Context

...ject proposals for detection purpose. Salient object detection models try to detect the most attention-grabbing object in a scene, and then segment the whole extent of that object [5, 40]. Liu et al. =-=[41]-=- combined local, regional, and global saliency measurements in a CRF framework. Achanta et al. [1] localized salient regions using a frequency-tuned approach. Cheng et al. [11, 14] proposed a salient ...

Citation Context

...d is 1,000+ times faster than most popular alternatives [3, 22, 48] (see Sec. 4). In addition, for efficient sliding window object detection, keeping the computational cost feasible is very important =-=[43, 51]-=-. Lampert et al. [39] presented an elegant branch-and-bound scheme for detection. However, it can only be used to speed up classifiers that users can provide a good bound on highest score. Also, some ...

Citation Context

...than most popular alternatives [3, 22, 48] (see Sec. 4). In addition, for efficient sliding window object detection, keeping the computational cost feasible is very important [43, 51]. Lampert et al. =-=[39]-=- presented an elegant branch-and-bound scheme for detection. However, it can only be used to speed up classifiers that users can provide a good bound on highest score. Also, some other efficient class...

Citation Context

...ic detectors to greatly reduce the computation for each of them. To the best of our knowledge, no prior method can satisfy all these ambitious goals simultaneously. Research from cognitive psychology =-=[47, 54]-=- and neurobiology [20,38] suggest that humans have a strong ability to perceive objects before identifying them. Based on the human reaction time that is observed and the biological signal transmissio...

Citation Context

...ion-grabbing object in a scene, and then segment the whole extent of that object [5, 40]. Liu et al. [41] combined local, regional, and global saliency measurements in a CRF framework. Achanta et al. =-=[1]-=- localized salient regions using a frequency-tuned approach. Cheng et al. [11, 14] proposed a salient object detection and segmentation method based on region contrast analysis and iterative graph bas...

Citation Context

...ect [5, 40]. Liu et al. [41] combined local, regional, and global saliency measurements in a CRF framework. Achanta et al. [1] localized salient regions using a frequency-tuned approach. Cheng et al. =-=[11, 14]-=- proposed a salient object detection and segmentation method based on region contrast analysis and iterative graph based segmentation. More recent research also tried to produce high quality saliency ...

Citation Context

...ate visual system, Itti et al. [36] proposed one of the first computational models for saliency detection, which estimates center-surrounded difference across multi-scale image features. Ma and Zhang =-=[42]-=- proposed a fuzzy growing model to analyze local contrast based saliency. Harel et al. [29] proposed normalizing center-surrounded feature maps for highlighting conspicuous parts. Although fixation po...

Citation Context

...iding window fashion [17, 25]. In order to reduce the number of windows each classifier needs to consider, training an objectness measure which is generic over categories has recently becomes popular =-=[2,3,21,22,48,49, 57]-=-. Objectness is usually represented as a value which reflects how likely an image window covers an object of any category [3]. A generic objectness measure has great potential to be used in a pre-filt...

Citation Context

...iding window fashion [17, 25]. In order to reduce the number of windows each classifier needs to consider, training an objectness measure which is generic over categories has recently becomes popular =-=[2,3,21,22,48,49, 57]-=-. Objectness is usually represented as a value which reflects how likely an image window covers an object of any category [3]. A generic objectness measure has great potential to be used in a pre-filt...

Citation Context

...d powerful feature “BING” to help the search for objects using objectness scores. Our work is motivated by the fact that objects are stand-alone things with well-defined closed boundaries and centers =-=[3, 26, 32]-=-. We observe that generic objects with well-defined closed boundaries share surprisingly strong correlation when looking at the norm of the gradient (see Fig. 1 and Sec. 3), after resizing of their co...

Citation Context

...iding window fashion [17, 25]. In order to reduce the number of windows each classifier needs to consider, training an objectness measure which is generic over categories has recently becomes popular =-=[2,3,21,22,48,49, 57]-=-. Objectness is usually represented as a value which reflects how likely an image window covers an object of any category [3]. A generic objectness measure has great potential to be used in a pre-filt...

Citation Context

...iding window fashion [17, 25]. In order to reduce the number of windows each classifier needs to consider, training an objectness measure which is generic over categories has recently becomes popular =-=[2,3,21,22,48,49, 57]-=-. Objectness is usually represented as a value which reflects how likely an image window covers an object of any category [3]. A generic objectness measure has great potential to be used in a pre-filt...

Citation Context

...void making decisions early on, by proposing a small number (e.g. 1,000) of category-independent proposals, that are expected to cover all objects in an image [3,22,48]. Producing rough segmentations =-=[6,21]-=- as object proposals has been shown to be an effective way of reducing search spaces for categoryspecific classifiers, whilst allowing the usage of strong classifiers to improve accuracy. However, the...

Citation Context

...ed great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications =-=[7, 8, 16, 31, 34, 35]-=- by automatically selecting good results [10,11]. However, these approaches are less likely to work for complicated images where many objects are presented and they are rarely dominant (e.g. VOC [23])...

Citation Context

...eves 96.2% DR using 1,000 proposals, and 99.5% DR using 5,000 proposals. The 3 methods [3, 48, 57] have been evaluated on the same benchmark and shown to outperform other alternative proposal methods =-=[6,21,25,30,50]-=-, saliency measures [33, 36], interesting point detectors [44], and HOG detector [17] (see [3] for the comparisons). Best viewed in color. are not used for training. Specifically, we train our method ...

Citation Context

... and segmentation method based on region contrast analysis and iterative graph based segmentation. More recent research also tried to produce high quality saliency maps in a filtering based framework =-=[46]-=-, using efficient data representation [12], or consider hierarchical structures [55]. Such salient object segmentation for simple images achieved great success in image scene analysis [15, 58], conten...

Citation Context

...d powerful feature “BING” to help the search for objects using objectness scores. Our work is motivated by the fact that objects are stand-alone things with well-defined closed boundaries and centers =-=[3, 26, 32]-=-. We observe that generic objects with well-defined closed boundaries share surprisingly strong correlation when looking at the norm of the gradient (see Fig. 1 and Sec. 3), after resizing of their co...

Citation Context

...le for generating object proposals for detection purpose. Salient object detection models try to detect the most attention-grabbing object in a scene, and then segment the whole extent of that object =-=[5, 40]-=-. Liu et al. [41] combined local, regional, and global saliency measurements in a CRF framework. Achanta et al. [1] localized salient regions using a frequency-tuned approach. Cheng et al. [11, 14] pr...

Citation Context

...d with [18] to enable realtime detection of thousands of object categories on a single machine. The efficiency of our method solves the efficiency bottleneck of proposal based object detection method =-=[53]-=-, possibly enabling realtime high quality object detection. We have demonstrated how to get a small set (e.g. 1,000) of proposals to cover nearly all (e.g. 96.2%) potential object regions, using very ...

Citation Context

...ic detectors to greatly reduce the computation for each of them. To the best of our knowledge, no prior method can satisfy all these ambitious goals simultaneously. Research from cognitive psychology =-=[47, 54]-=- and neurobiology [20,38] suggest that humans have a strong ability to perceive objects before identifying them. Based on the human reaction time that is observed and the biological signal transmissio...

Citation Context

... representation [12], or consider hierarchical structures [55]. Such salient object segmentation for simple images achieved great success in image scene analysis [15, 58], content aware image editing =-=[13, 56, 60]-=-, and it can be used as a cheap tool to process large number of Internet images or build robust applications [7, 8, 16, 31, 34, 35] by automatically selecting good results [10,11]. However, these appr...

Citation Context

... representation [12], or consider hierarchical structures [55]. Such salient object segmentation for simple images achieved great success in image scene analysis [15, 58], content aware image editing =-=[13, 56, 60]-=-, and it can be used as a cheap tool to process large number of Internet images or build robust applications [7, 8, 16, 31, 34, 35] by automatically selecting good results [10,11]. However, these appr...

Citation Context

...ented by BING and BING-generic, respectively. As we see, the behavior of these two curves are almost identical, which demonstrates the generalize ability of our proposals. Notice that the recent work =-=[18]-=- enables 20 seconds testing time for detecting 100, 000 object classes, by reducing the computational complexity of traditional multi-class detection from O(LC) to O(L), where L is the number of locat...

Citation Context

... segmentation. More recent research also tried to produce high quality saliency maps in a filtering based framework [46], using efficient data representation [12], or consider hierarchical structures =-=[55]-=-. Such salient object segmentation for simple images achieved great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it can be used as a cheap tool to process la...

Citation Context

...d research into three categories: fixation prediction, salient object detection, and objectness proposal generation. Fixation prediction models aim at predicting saliency points of human eye movement =-=[4, 37]-=-. Inspired by neurobiology research about early primate visual system, Itti et al. [36] proposed one of the first computational models for saliency detection, which estimates center-surrounded differe...

Citation Context

... representation [12], or consider hierarchical structures [55]. Such salient object segmentation for simple images achieved great success in image scene analysis [15, 58], content aware image editing =-=[13, 56, 60]-=-, and it can be used as a cheap tool to process large number of Internet images or build robust applications [7, 8, 16, 31, 34, 35] by automatically selecting good results [10,11]. However, these appr...

Citation Context

...ed great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications =-=[7, 8, 16, 31, 34, 35]-=- by automatically selecting good results [10,11]. However, these approaches are less likely to work for complicated images where many objects are presented and they are rarely dominant (e.g. VOC [23])...

Citation Context

...ed great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications =-=[7, 8, 16, 31, 34, 35]-=- by automatically selecting good results [10,11]. However, these approaches are less likely to work for complicated images where many objects are presented and they are rarely dominant (e.g. VOC [23])...

Citation Context

...ontrast analysis and iterative graph based segmentation. More recent research also tried to produce high quality saliency maps in a filtering based framework [46], using efficient data representation =-=[12]-=-, or consider hierarchical structures [55]. Such salient object segmentation for simple images achieved great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it...

Citation Context

...ed great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications =-=[7, 8, 16, 31, 34, 35]-=- by automatically selecting good results [10,11]. However, these approaches are less likely to work for complicated images where many objects are presented and they are rarely dominant (e.g. VOC [23])...

Citation Context

...iding window fashion [17, 25]. In order to reduce the number of windows each classifier needs to consider, training an objectness measure which is generic over categories has recently becomes popular =-=[2,3,21,22,48,49, 57]-=-. Objectness is usually represented as a value which reflects how likely an image window covers an object of any category [3]. A generic objectness measure has great potential to be used in a pre-filt...

Citation Context

...ed great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications =-=[7, 8, 16, 31, 34, 35]-=- by automatically selecting good results [10,11]. However, these approaches are less likely to work for complicated images where many objects are presented and they are rarely dominant (e.g. VOC [23])...

Citation Context

...ed great success in image scene analysis [15, 58], content aware image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications =-=[7, 8, 16, 31, 34, 35]-=- by automatically selecting good results [10,11]. However, these approaches are less likely to work for complicated images where many objects are presented and they are rarely dominant (e.g. VOC [23])...

Citation Context

...le for generating object proposals for detection purpose. Salient object detection models try to detect the most attention-grabbing object in a scene, and then segment the whole extent of that object =-=[5, 40]-=-. Liu et al. [41] combined local, regional, and global saliency measurements in a CRF framework. Achanta et al. [1] localized salient regions using a frequency-tuned approach. Cheng et al. [11, 14] pr...

Citation Context

...ect [5, 40]. Liu et al. [41] combined local, regional, and global saliency measurements in a CRF framework. Achanta et al. [1] localized salient regions using a frequency-tuned approach. Cheng et al. =-=[11, 14]-=- proposed a salient object detection and segmentation method based on region contrast analysis and iterative graph based segmentation. More recent research also tried to produce high quality saliency ...

Citation Context

...e image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications [7, 8, 16, 31, 34, 35] by automatically selecting good results =-=[10,11]-=-. However, these approaches are less likely to work for complicated images where many objects are presented and they are rarely dominant (e.g. VOC [23]). Objectness proposal generation methods avoid m...

Citation Context

...ng very simple BING features. It would be interesting to introduce other additional cues to further reduce the number of proposals while maintaining high detection rate, and explore more applications =-=[9]-=- using BING. To encourage future works, we make the source code, links to related methods, FAQs, and live discussions available in the project page: http://mmcheng.net/bing/. Acknowledges: We acknowle...

Citation Context

...tations (see Sec. 4 for evaluation criteria). Discussion. As illustrated in Fig. 1d, the learned linear model w (see Sec. 4 for experimental settings), looks simAlgorithm 1 Binary approximate model w =-=[28]-=-. Input: w, Nw Output: {βj}Nwj=1, {aj}Nwj=1 Initialize residual: ε = w for j = 1 to Nw do aj =sign(ε) βj = 〈aj , ε〉/‖aj‖2 (project ε onto aj) ε← ε− βjaj (update residual) end for ilar to the multi-siz...

Citation Context

...d framework [46], using efficient data representation [12], or consider hierarchical structures [55]. Such salient object segmentation for simple images achieved great success in image scene analysis =-=[15, 58]-=-, content aware image editing [13, 56, 60], and it can be used as a cheap tool to process large number of Internet images or build robust applications [7, 8, 16, 31, 34, 35] by automatically selecting...

Citation Context

...s [3, 57] and classic sliding window based object detection methods [17, 25], For some object categories, a bounding box might not localize the object instances as accurately as a segmentation region =-=[6,21,22,45]-=-, e.g. a snake, wires, etc. Future works. The high quality and efficiency of our method make it suitable for realtime multi-category object detection applications and large scale image collections (e....

Citation Context

...occluded than upper parts. This is represented by w placing less confidence in the lower regions. 3.3. Binarized normed gradients (BING) To make use of recent advantages in model binary approximation =-=[28, 59]-=-, we propose an accelerated version of NG feature, namely binarized normed gradients (BING), to speed up the feature extraction and testing process. Our learned linear model w ∈ R64 can be approximate...