Shouchuan Wu, Haitao Zhao, Shaoyuan Sun. Depth Estimation from Monocular Infrared Video Based on Bi-Recursive Convolutional Neural Network[J]. Acta Optica Sinica, 2017, 37(12): 1215003
- Acta Optica Sinica
- Vol. 37, Issue 12, 1215003 (2017)
Fig. 1. Mechanism of bi-recursive convolution
Fig. 2. Diagram for BrCNN
Fig. 3. Effect of dividing ground truth depth of infrared video into different depth levels. (a) Infrared video; (b) ground truth; (c) 10 layers; (d) 20 layers; (e) 30 layers
Fig. 4. Output of different bi-recursive convolutional layers. (a) Infrared video; (b) output of the first bi-recursive convolutional layer; (c) output of the second bi-recursive convolutional layer; (d) output of the third bi-recursive convolutional layer
Fig. 5. Depth estimation results of different models. (a) Infrared video; (b) ground truth depth; (c) depth estimated by BrCNN; (d) depth estimated by AlexNet; (e) depth estimated by VGG16; (f) depth estimated by Res34
Fig. 6. (a) Mean relative error of BrCNN and VGG16 model; (b) 2-norm of difference between adjacent frames
Fig. 7. Depth estimation results of larger 2-norm between frame differences. (a) Continuous three frames of infrared video; (b) estimation results of BrCNN
| |||||||||||||||||||||||||||||||||||||||||||||||
Table 1. Comparison of depth estimation results between BrCNN model and other classical network models
Set citation alerts for the article
Please enter your email address
© Copyright 2018-2021 | Chinese Laser Press. All Rights Reserved 沪ICP备15018463号-20