In Defense of Classical Image Processing: Fast Depth Completion on the CPU

By Jason Ku, Ali Harakeh, Steven L. Waslander. Canadian Conference on Computer and Robot Vision, 2018

337 citations recorded by the source corpus.

With the rise of data driven deep neural networks as a realization of universal function approximators, most research on computer vision problems has moved away from handcrafted classical image processing algorithms. This paper shows that with a well designed algorithm, we are capable of outperforming neural network based methods on the task of depth completion. The proposed algorithm is simple and fast, runs on the CPU, and relies only on basic image processing operations to perform depth completion of sparse LIDAR depth data. We evaluate our algorithm on the challenging KITTI depth completion benchmark, and at the time of submission, our method ranks first on the KITTI test server among all published methods. Furthermore, our algorithm is data independent, requiring no training data to perform the task at hand. The code written in Python is publicly available at https://github.com/kujason/ip_basic

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/260eff563879654caa99fc599cfbdca4617d291a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

PVN3D: A Deep Point-Wise 3D Keypoints Voting Network for 6DoF Pose Estimation (2019, methodology): “A hole completion al-gorithm [25] is also applied to improve the quality of depth images.”
Categorical Depth Distribution Network for Monocular 3D Object Detection (2021, background): “Depth completion [20] is performed to generate depth values at each pixel in the image.”
FFB6D: A Full Flow Bidirectional Fusion Network for 6D Pose Estimation (2021, positive): “Synthetic images were also taken for training as in [68] and the hole completion algorithm [30] is applied for hole filling for depth images as in [17].”
Learning Guided Convolutional Network for Depth Completion (2019, methodology): “These methods have achieved signiﬁcantly improvements over conventional approaches [8]– [10].”
Self-Supervised Sparse-to-Dense: Self-Supervised Depth Completion from LiDAR and Monocular Camera (2018, neutral): “[20] developed a simple and fast interpolation-based algorithm that runs on CPUs.”