A Novel Multi-layer Framework for Tiny Obstacle Discovery

Abstract

For tiny obstacle discovery in a monocular image, edge is a fundamental visual element. Nevertheless, because of various reasons, e.g., noise and similar color distribution with background, it is still difficult to detect the edges of tiny obstacles at long distance. In this paper, we propose an obstacleaware discovery method to recover the missing contours of these obstacles, which helps to obtain obstacle proposals as much as possible. First, by using visual cues in monocular images, several multi-layer regions are elaborately inferred to reveal the distances from the camera. Second, several novel obstacleaware occlusion edge maps are constructed to well capture the contours of tiny obstacles, which combines cues from each layer. Third, to ensure the existence of the tiny obstacle proposals, the maps from all layers are used for proposals extraction. Finally, based on these proposals containing tiny obstacles, a novel obstacle-aware regressor is proposed to generate an obstacle occupied probability map with high confidence. The convincing experimental results with comparisons on the Lost and Found dataset demonstrate the effectiveness of our approach, achieving around 9.5% improvement on the accuracy than FPHT and PHT, it even gets comparable performance to MergeNet. Moreover, our method outperforms the state-of-theart algorithms and significantly improves the discovery ability for tiny obstacles at long distance.

Tiny Obstacle Discovery by Occlusion-Aware Multilayer Regression

“…In this way, the pyramid bottom, i.e., the enhanced edge probability map Ê, is generated. Unlike [25], only the bottom map is reserved in this paper, because this map Ê expresses the obstacle contour more clearly than others. The 1st rows of Fig.…”

“…Moreover, the OBDR further raises the confidence of the obstacle proposal. Thus, the fused predictions better represent the obstacles than our base approach [25]. 5) Obstacle-Occupied Probability Map: The scores of the top 50% proposals in B I are accumulated in the corresponding pixels to produce a probability map P :…”

“…As the best variant, the same sampling parameter is employed in the next comparisons. 6) Evaluation of the enhancement to our base algorithm: To evaluate the enhancement to our base algorithm [25], Fig. 10 and Table IV depict the results of three variants, i.e., OP+ORDR@4, BP+ORDR@4, and BP+Fusion@4.…”

“…II. RELATED WORK An earlier version of this work appeared in [25]. Since our method closely depends on edge boxes [26] and object-level proposals [20], they are briefly introduced below.…”

Occlusion-Shared and Feature-Separated Network for Occlusion Relationship Reasoning

“…Reasoning the occlusion relationship of objects from monocular image is fundamental in computer vision and mobile robot applications, such as [11,2,24,17,29]. Furthermore, it can be regarded as crucial elements for scene understanding and visual perception [40,42,43,39,18], such as object detection, image segmentation and 3D reconstruction [6,1,37,7,26,34]. From the perspective of * Corresponding Author Occlusion relationship (the red arrows) is represented by orientation θ ∈ (−π, π] (tangent direction of the edge), using the "left" rule where the left side of the arrow means foreground area.…”