--- title: IEBins Depth Perception emoji: 🏢 colorFrom: pink colorTo: pink sdk: gradio sdk_version: 4.15.0 app_file: app.py pinned: false license: mit ---

IEBins: Iterative Elastic Bins for Monocular Depth Estimation

Shuwei Shao1  Zhongcai Pei1  Xingming Wu1  Zhong Liu1  Weihai Chen2  Zhengguo Li3
1Beihang University, 2Anhui University, 3A*STAR

• NeurIPS 2023 •

[![KITTI Benchmark](https://img.shields.io/badge/KITTI%20Benchmark-2nd%20among%20all%20at%20submission%20time-blue)](https://www.cvlibs.net/datasets/kitti/eval_depth.php?benchmark=depth_prediction) [![Hugging Space Badge](https://img.shields.io/badge/🤗-Open%20In%20Spaces-blue.svg)](umuthopeyildirim/IEBins-Depth-Perception) ## Abstract
We propose a novel concept of iterative elastic bins for the classification-regression-based MDE. The proposed IEBins aims to search for high-quality depth by progressively optimizing the search range, which involves multiple stages and each stage performs a finer-grained depth search in the target bin on top of its previous stage. To alleviate the possible error accumulation during the iterative process, we utilize a novel elastic target bin to replace the original target bin, the width of which is adjusted elastically based on the depth uncertainty. --- ## Installation ``` conda create -n iebins python=3.8 conda activate iebins conda install pytorch=1.10.0 torchvision cudatoolkit=11.1 pip install matplotlib, tqdm, tensorboardX, timm, mmcv, open3d ``` ## Datasets You can prepare the datasets KITTI and NYUv2 according to [here](https://github.com/cleinc/bts/tree/master/pytorch) and download the SUN RGB-D dataset from [here](https://rgbd.cs.princeton.edu/), and then modify the data path in the config files to your dataset locations. ## Training First download the pretrained encoder backbone from [here](https://github.com/microsoft/Swin-Transformer), and then modify the pretrain path in the config files. If you want to train the KITTI_Official model, first download the pretrained encoder backbone from [here](https://drive.google.com/file/d/1qjDnMwmEz0k0XWh7GP2aNPGiAjvOPF_5/view?usp=drive_link), which is provided by [MIM](https://github.com/SwinTransformer/MIM-Depth-Estimation). Training the NYUv2 model: ``` python iebins/train.py configs/arguments_train_nyu.txt ``` Training the KITTI_Eigen model: ``` python iebins/train.py configs/arguments_train_kittieigen.txt ``` Training the KITTI_Official model: ``` python iebins_kittiofficial/train.py configs/arguments_train_kittiofficial.txt ``` ## Evaluation Evaluate the NYUv2 model: ``` python iebins/eval.py configs/arguments_eval_nyu.txt ``` Evaluate the NYUv2 model on the SUN RGB-D dataset: ``` python iebins/eval_sun.py configs/arguments_eval_sun.txt ``` Evaluate the KITTI_Eigen model: ``` python iebins/eval.py configs/arguments_eval_kittieigen.txt ``` To generate KITTI Online evaluation data for the KITTI_Official model: ``` python iebins_kittiofficial/test.py --data_path path to dataset --filenames_file ./data_splits/kitti_official_test.txt --max_depth 80 --checkpoint_path path to pretrained checkpoint --dataset kitti --do_kb_crop ``` ## Qualitative Depth and Point Cloud Results You can download the qualitative depth results of [IEBins](https://arxiv.org/abs/2309.14137), [NDDepth](https://arxiv.org/abs/2309.10592), [NeWCRFs](https://openaccess.thecvf.com/content/CVPR2022/html/Yuan_Neural_Window_Fully-Connected_CRFs_for_Monocular_Depth_Estimation_CVPR_2022_paper.html), [PixelFormer](https://openaccess.thecvf.com/content/WACV2023/html/Agarwal_Attention_Attention_Everywhere_Monocular_Depth_Prediction_With_Skip_Attention_WACV_2023_paper.html), [AdaBins](https://openaccess.thecvf.com/content/CVPR2021/html/Bhat_AdaBins_Depth_Estimation_Using_Adaptive_Bins_CVPR_2021_paper.html) and [BTS](https://arxiv.org/abs/1907.10326) on the test sets of NYUv2 and KITTI_Eigen from [here](https://pan.baidu.com/s/1zaFe40mwpQ5cvdDlLZRrCQ?pwd=vfxd) and download the qualitative point cloud results of IEBins, NDDepth, NeWCRFS, PixelFormer, AdaBins and BTS on the NYUv2 test set from [here](https://pan.baidu.com/s/1WwpFuPBGBUaSGPEdThJ6Rw?pwd=n9rw). If you want to derive these results by yourself, please refer to the test.py. If you want to perform inference on a single image, run: ``` python iebins/inference_single_image.py --dataset kitti or nyu --image_path path to image --checkpoint_path path to pretrained checkpoint --max_depth 80 or 10 ``` Then you can acquire the qualitative depth result. ## Models | Model | Abs Rel | Sq Rel | RMSE | a1 | a2 | a3 | Link | | -------------------- | :-----: | :----: | :---: | :---: | :---: | :---: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------: | | NYUv2 (Swin-L) | 0.087 | 0.040 | 0.314 | 0.936 | 0.992 | 0.998 | [[Google]](https://drive.google.com/file/d/14Rn-vxvpXO2EXRaWqCPmh2JufvOurwtl/view?usp=drive_link) [[Baidu]](https://pan.baidu.com/s/1E2KAHtQ-ul99RGp_G7QK1w?pwd=7o4d) | | NYUv2 (Swin-T) | 0.108 | 0.061 | 0.375 | 0.893 | 0.984 | 0.996 | [[Google]](https://drive.google.com/file/d/1eYkTb3grbDitQ9tJdg1DhAOaGmqgHWHK/view?usp=drive_link) [[Baidu]](https://pan.baidu.com/s/1v5_MJtP0YOSoark9Yw1RaQ?pwd=2k5d) | | KITTI_Eigen (Swin-L) | 0.050 | 0.142 | 2.011 | 0.978 | 0.998 | 0.999 | [[Google]](https://drive.google.com/file/d/1xaVLDq7zJ-C2GtFvABolSUtK7gzvNQNd/view?usp=drive_link) [[Baidu]](https://pan.baidu.com/s/16mRrKrr9PdZhuZ3ZlkmNlA?pwd=lcjd) | | KITTI_Eigen (Swin-T) | 0.056 | 0.169 | 2.205 | 0.970 | 0.996 | 0.999 | [[Google]](https://drive.google.com/file/d/1s0LXZmS6_Q4_H_0hmbOldPcVhlRw8Dut/view?usp=drive_link) [[Baidu]](https://pan.baidu.com/s/1xgeqIX5WP5F2MFwypMWV5A?pwd=ygfi) | | Model | SILog | Abs Rel | Sq Rel | RMSE | a1 | a2 | a3 | Link | | ------------------------- | :---: | :-----: | :----: | :--: | :---: | :---: | :---: | :-----------------------------------------------------------------------------------------------: | | KITTI_Official (Swinv2-L) | 7.48 | 5.20 | 0.79 | 2.34 | 0.974 | 0.996 | 0.999 | [[Google]](https://drive.google.com/file/d/19ARBiDTIvtZSWJVvhbEWBcZMonXsiOX1/view?usp=drive_link) | ## Citation If you find our work useful in your research please consider citing our paper: ``` @inproceedings{shao2023IEBins, title={IEBins: Iterative Elastic Bins for Monocular Depth Estimation}, author={Shao, Shuwei and Pei, Zhongcai and Wu, Xingming and Liu, Zhong and Chen, Weihai and Li, Zhengguo}, booktitle={Advances in Neural Information Processing Systems (NeurIPS)}, year={2023} } ``` ## Contact If you have any questions, please feel free to contact swshao@buaa.edu.cn. ## Acknowledgement Our code is based on the implementation of [NeWCRFs](https://github.com/aliyun/NeWCRFs) and [BTS](https://github.com/cleinc/bts). We thank their excellent works.