Deep Embedding Features for Salient Object Detection

Abstract

Benefiting from the rapid development of Convolutional Neural Networks (CNNs), some salient object detection methods have achieved remarkable results by utilizing multi-level convolutional features. However, the saliency training datasets is of limited scale due to the high cost of pixel-level labeling, which leads to a limited generalization of the trained model on new scenarios during testing. Besides, some FCN-based methods directly integrate multi-level features, ignoring the fact that the noise in some features are harmful to saliency detection. In this paper, we propose a novel approach that transforms prior information into an embedding space to select attentive features and filter out outliers for salient object detection. Our network firstly generates a coarse prediction map through an encorder-decorder structure. Then a Feature Embedding Network (FEN) is trained to embed each pixel of the coarse map into a metric space, which incorporates much attentive features that highlight salient regions and suppress the response of non-salient regions. Further, the embedded features are refined through a deep-to-shallow Recursive Feature Integration Network (RFIN) to improve the details of prediction maps. Moreover, to alleviate the blurred boundaries, we propose a Guided Filter Refinement Network (GFRN) to jointly optimize the predicted results and the learnable guidance maps. Extensive experiments on five benchmark datasets demonstrate that our method outperforms state-of-the-art results. Our proposed method is end-to-end and achieves a realtime speed of 38 FPS.

Progressive Feature Polishing Network for Salient Object Detection

“…In contrast to that, our method polishes the representations at every level with multi-level context information at each step. Moreover, many methods utilize an extra refinement module, either as a part of their model or as a post-process, to further recover the details of the predicted results, such as DenseCRF (Hou et al 2017;Liu, Han, and Yang 2018), BRN (Wang et al 2018) and GFRN (Zhuge, Zeng, and Lu 2019). In contrast, our method delivers superior performance without such modules.…”

“…Wang (Wang et al 2018) propose to better localize salient objects by exploiting contextual information with attentive mechanism. Zhuge (Zhuge, Zeng, and Lu 2019) employ a structure which embeds prior information to generate attentive features and filter out cluttered information. Different from above methods which design sophisticated structure to make information fusion, we use simple structure to polish multi-level features in recurrent manner and in parallel.…”

“…We use the cross-entropy loss between the final predicted saliency map and ground truth to train our model end-toend. Following previous works (Hou et al 2017;Liu, Han, and Yang 2018;Zhuge, Zeng, and Lu 2019), side outputs are also employed to calculate auxiliary losses. In detail, 1x1 convolutional layers are performed on the multi-level feature maps before the Fusion Module to obtain a series of intermediate results.…”

Multi-Type Self-Attention Guided Degraded Saliency Detection

“…Liu et al (2019) built two pooling-based modules, one module provided different layers the location information of potential salient objects, while the other merged features. Zhuge, Zeng, and Lu (2019) transformed prior information into an embedding space to select attentive features and to filter out outliers. Feng, Lu, and Ding (2019) proposed a boundary-enhanced loss for learning exquisite object boundaries.…”

Learning Deep Relations to Promote Saliency Detection

“…RADF (Hu et al 2018) use recurrently aggregated deep features to detect saliency object. Zhuge et al (Zhuge, Zeng, and Lu 2019) argue that the noise in some features are harmful to saliency detection. PiCANet (Liu, Han, and Yang 2018), RAS (Chen et al 2018), and PFA (Zhao and Wu 2019)) both adopt the attention mechanism to get better saliency result.…”