CAGNet: Content-Aware Guidance for Salient Object Detection

Abstract

Beneficial from Fully Convolutional Neural Networks (FCNs), saliency detection methods have achieved promising results. However, it is still challenging to learn effective features for detecting salient objects in complicated scenarios, in which i) non-salient regions may have "salient-like" appearance; ii) the salient objects may have differentlooking regions. To handle these complex scenarios, we propose a Feature Guide Network which exploits the nature of low-level and high-level features to i) make foreground and background regions more distinct and suppress the non-salient regions which have "salient-like" appearance; ii) assign foreground label to different-looking salient regions.Furthermore, we utilize a Multi-scale Feature Extraction Module (MFEM) for each level of abstraction to obtain multi-scale contextual information. Finally, we design a loss function which outperforms the widely-used Cross-entropy loss. By adopting four different pre-trained models as the backbone, we prove that our method is very general with respect to the choice of the backbone model. Experiments on five challenging datasets demonstrate that our method achieves the state-of-the-art performance in terms of different evaluation metrics. Additionally, our approach contains fewer parameters than the existing ones, does not need any post-processing, and runs fast at a real-time speed of 28 FPS when processing a 480 × 480 image.

Contour-Aware Recurrent Cross Constraint Network for Salient Object Detection

“…We compare the proposed CARCCNet against 10 other stateof-the-art saliency models: Amulet [44], NLDF [45], C2SNet [20], R3Net [17], DGRL [28], CapSal [46], HRS-D [47], CAGNet [48], ScriSal [49], and EDNS [50]. The SOD maps of these extant methods are provided by the authors or computed by their released codes.…”

Multiscale Deep Network with Centerness-Aware Loss for Salient Object Detection

“…Particularly, fully convolutional networks (FCN) show their advantage and refresh the state-of-theart records in saliency prediction task. e encoder-decoder framework is frequently used in the FCN-based saliency models [3,[15][16][17][18][19][38][39][40]. Liu et al [16] proposed a novel network to embed local and global pixelwise contextual attention modules into a U-shape network.…”

Auto-MSFNet: Search Multi-scale Fusion Network for Salient Object Detection

“…The performance of the proposed method is compared with 13 state-of-the-art salient object detection methods including Amulet [57], DSS [20], C2S [26], AFnet [18], Poolnet [30], EGnet [60], Gatenet [62], CAGnet [33], ISTDnet [63], MInet [34], SCRnet [47], F3net [43] and LDnet [44]. For a fair comparison, we use two corresponding backbones to compare 11 methods based on VGG-16 [40] and 8 methods based on ResNet-50 [19].…”