Dynamo-Depth: Fixing Unsupervised Depth Estimation for Dynamical Scenes
Authors: Yihong Sun, Bharath Hariharan
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 5 Experiments, Specifically, we evaluate on three datasets Waymo Open [34], nu Scenes [3], and KITTI [9] with Eigen split [6]. As shown in Table 1, our proposed approach outperforms prior arts on both nu Scenes and Waymo Open across all metrics, with over 57% and 21% reduction in overall Abs Rel for nu Scenes and Waymo Open, respectively. |
| Researcher Affiliation | Academia | Yihong Sun Cornell University yihong@cs.cornell.edu Bharath Hariharan Cornell University bharathh@cs.cornell.edu |
| Pseudocode | No | No pseudocode or clearly labeled algorithm block was found in the paper. |
| Open Source Code | Yes | Code and additional results are available at https://dynamo-depth.github.io. |
| Open Datasets | Yes | Specifically, we evaluate on three datasets Waymo Open [34], nu Scenes [3], and KITTI [9] with Eigen split [6]. |
| Dataset Splits | Yes | For the Waymo Open Dataset [34], 76,852 front camera image-triplets from the provided train set containing 798 video sequences are used for training while 2,216 front camera images uniformly sampled from the provided validation set containing 202 video sequences are used for evaluation. |
| Hardware Specification | Yes | The proposed method is trained on four NVIDIA 2080 Ti with a total batch size of 12 and an epoch size of 8000 sampled batches. |
| Software Dependencies | No | The paper mentions software components like 'Adam optimizer', 'ResNet18', 'Monodepth2', and 'Lite Mono' by name, but does not provide specific version numbers for any of them or for underlying programming languages/frameworks. |
| Experiment Setup | Yes | Adam optimizer [20] is used with an initial learning rate of 5e-5 and drops to 2.5e-5 after 10 epochs. Motion Initialization lasts 5 epochs and takes place after the depth network and complete flow network have been trained for 1 epoch each. After initialization, the system is trained for 20 epochs, totalling approximately 20 hours. The hyperparameter values are the same for all experiments and are provided in Appendix B.1, along with the rest of the details of the model architecture. |