Uncertainty-aware Joint Salient Object and Camouflaged Object Detection

Abstract

Salient objects attract human attention and usually stand out clearly from their surroundings. In contrast, camouflaged objects share similar colors or textures with the environment. In this case, salient objects are typically non-camouflaged, and camouflaged objects are usually not salient. Due to this inherent "contradictory" attribute, we introduce an uncertainty-aware learning pipeline to extensively explore the contradictory information of salient object detection (SOD) and camouflaged object detection (COD) via datalevel and task-wise contradiction modeling. We first exploit the "dataset correlation" of these two tasks and claim that the easy samples in the COD dataset can serve as hard samples for SOD to improve the robustness of the SOD model. Based on the assumption that these two models should lead to activation maps highlighting different regions of the same input image, we further introduce a "contrastive" module with a joint-task contrastive learning framework to explicitly model the contradictory attributes of these two tasks. Different from conventional intra-task contrastive learning for unsupervised representation learning, our "contrastive" module is designed to model the task-wise correlation, leading to cross-task representation learning. To better understand the two tasks from the perspective of uncertainty, we extensively investigate the uncertainty estimation techniques for modeling the main uncertainties of the two tasks, namely "task uncertainty" (for SOD) and "data uncertainty" (for COD), and aiming to effectively estimate the challenging regions for each task to achieve difficulty-aware learning. Experimental results on benchmark datasets demonstrate that our solution leads to both state-of-the-art performance and informative uncertainty estimation.

Towards Deeper Understanding of Camouflaged Object Detection

“…that SINet-V2 [80] achieves the best performance in Background Matching, Complex Background and Corner Position and Disruptive Coloration attributes. UJCS [21] performs relatively superior in Background Matching, Mimicry (best) and Salient Object. The performance of C2FNet [77] ranks the top three and it excels in Small Object.…”

“…where the latent variables within the latent variable models aim to represent the "inherent randomness" of the task. Note that L2 loss function is used in both models for the generator, and we adopt the GAN based framework as in [21] and the VAE based framework as in [100]. The performance of each model is shown as "GAN" and "VAE" respectively in Table 7.…”

“…In recent years, many studies take advantage of the powerful feature representation ability of deep learning to construct structure contrast between the object and the background [11], [12], [16]. Although similar methods via contrast modeling for COD have achieved some favourable performance [17], [18], [19], [20], [21], [22], it is still challenging because some detailed information that is necessary for fine-grained recognition and better understanding about camouflage may be ignored by the deep neural network, leading to less effective models.…”

“…The existing camouflaged object detection models [18], [19], [20], [21] are designed based on binary camouflage ground truth maps as shown in Fig. 1 ("Binary"), which only capture the scope of the camouflaged objects without fully exploring the basic mechanism of camouflage.…”

Deep Gradient Learning for Efficient Camouflaged Object Detection

“…Recent studies [11][12][13][14] present compelling results based on the supervision of the whole object-level ground-truth mask. Later, various sophisticated techniques, e.g., boundarybased [15][16][17] and uncertainty-guided [18,19], were developed to augment COD's underlying representations. However, features learned from boundary-supervised or uncertainty-based models usually respond to the sparse edge of camouflage objects, thereby introducing noisy features, especially for complex scenes (see Fig.…”