Scale-Aware Attention-Based PillarsNet (SAPN) Based 3D Object Detection for Point Cloud

Figure 6

Illustrations of the selected pseudo-images from different resolutions. (a) is the source image and (b) is the corresponding BEV image. (c), (d), and (e) are pseudo-images from high, middle, and low resolutions, respectively. (f), (g), and (h) are selected zoom-in regions that contain two car objects which are labeled with red rectangles.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)