Double Refinement Network for Efficient Monocular Depth Estimation

Monocular depth estimation is the task of obtaining a measure of distance for each pixel using a single image. It is an important problem in computer vision and is usually solved using neural networks. Though recent works in this area have shown significant improvement in accuracy, the state-of-the-art methods tend to require massive amounts of memory and time to process an image. The main purpose of this work is to improve the performance of the latest solutions with no decrease in accuracy. To this end, we introduce the Double Refinement Network architecture. The proposed method achieves state-of-the-art results on the standard benchmark RGB-D dataset NYU Depth v2, while its frames per second rate is significantly higher (up to 18 times speedup per image at batch size 1) and the RAM usage is lower.

Authors

• Nikita Durasov
• Mikhail Romanov
• Valeriya Bubnova
• Pavel Bogomolov
• Anton Konushin

Keywords

• Interpolation
• Accuracy
• Depth measurement
• Memory management
• Neural networks
• Random access memory
• Network architecture
• Real-time systems
• Iterative methods
• Standards
• Depth Estimation
• Double Network
• Monocular Depth Estimation
• Neural Network
• Computer Vision
• Decrease In Accuracy
• Computer Vision Problems
• Standard Depth
• Loss Function
• Root Mean Square Error
• Deep Neural Network
• Convolutional Layers
• Feature Maps
• Ordinal Regression
• Deep Convolutional Neural Network
• Focal Length
• Depth Map
• Pre-trained Network
• Bilinear Interpolation
• Dilated Convolution
• Intermediate Output
• Auxiliary Loss