SAL-Net: Self-Supervised Attribute Learning for Object Recognition and Segmentation

<div>The visualization of each module in the proposed multitarget recognition and segmentation system: (a) region proposals generated by selective search algorithm; (b) SAL-Net predictions filtered by score threshold 0.6; (c) results processed by Fusion NMS.</div>

Wireless Communications and Mobile Computing

fig7

Figure 7

Figure 7: SAL-Net: Self-Supervised Attribute Learning for Object Recognition and Segmentation 