Jian Li, Jie Tang, Ping Tan, Boshi An, Fei-Peng Tian
We lifted 6 functions out of this paper's own repositories and ran 1 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| kakaxi314/BP-Net | canonical | 1 of 6 |
| Function | Status | Where it lives |
|---|---|---|
| config_param | Ran | kakaxi314/BP-Net/utils.py code served (permissive licence) · get_code("1ee8a417a1da1f5c") |
| Conv1x1 | Not yet run | kakaxi314/BP-Net/models/utils.py code served (permissive licence) · get_code("c17d19aba45c1c21") |
| Conv3x3 | Not yet run | kakaxi314/BP-Net/models/utils.py code served (permissive licence) · get_code("f8ae19885bce102c") |
| add_pca_jitter | Not yet run | kakaxi314/BP-Net/augs.py code served (permissive licence) · get_code("9e00ec998b85a335") |
| compute_pca | Not yet run | kakaxi314/BP-Net/augs.py code served (permissive licence) · get_code("ccda8aad864983b8") |
| read_calib_file | Not yet run | kakaxi314/BP-Net/datasets/kitti.py code served (permissive licence) · get_code("3c91deaa8c673fd6") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Depth completion aims to derive a dense depth map from sparse depth measurements with a synchronized color image. Current state-of-the-art (SOTA) methods are predominantly propagation-based, which work as an iterative refinement on the initial estimated dense depth. However, the initial depth estimations mostly result from direct applications of convolutional layers on the sparse depth map. In this paper, we present a Bilateral Propagation Network (BP-Net), that propagates depth at the earliest stage to avoid directly convolving on sparse data. Specifically, our approach propagates the target depth from nearby depth measurements via a non-linear model, whose coefficients are generated through a multi-layer perceptron conditioned on both radiometric difference and spatial distance. By integrating bilateral propagation with multi-modal fusion and depth refinement in a multi-scale framework, our BP-Net demonstrates outstanding performance on both indoor and outdoor scenes. It achieves SOTA on the NYUv2 dataset and ranks 1st on the KITTI depth completion benchmark at the time of submission. Experimental results not only show the effectiveness of bilateral propagation but also emphasize the significance of early-stage propagation in contrast to the refinement stage. Our code and trained models will be available on the project page. * indicates the corresponding author. 1 The exact name should be image guided depth completion, if considering some early attempts don't utilize color images. Color Image Sparse Depth MF. Dense Depth (a) 1-stage depth completion (e.g. [16, 27, 41]). MF. Post. (b) 2-stage Depth Completion (e.g. [6-9]) Dense Depth (b) 2-stage depth completion (e.g. [3, 22, 31]). Pre. MF. Post. (c) 3-stage Depth Completion (our BP-Net) Dense Depth (c) 3-stage depth completion (our BP-Net).
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2403.11270")
get_code_for_paper("2403.11270")
have("2403.11270")
Connect an agent — have() is free.