--- task_categories: - object-detection license: mit tags: - computer vision - amodal-tracking - object-tracking - amodal-perception configs: - config_name: default data_files: - split: train path: "amodal_annotations/train.json" - split: validation path: "amodal_annotations/validation.json" - split: test path: "amodal_annotations/test.json" extra_gated_prompt: "To download the AVA and HACS videos you have to agree to terms and conditions." extra_gated_fields: You will use the Datasets only for non-commercial research and educational purposes.: type: select options: - Yes - No You will NOT distribute the Datasets or any parts thereof.: type: select options: - Yes - No Carnegie Mellon University makes no representations or warranties regarding the datasets, including but not limited to warranties of non-infringement or fitness for a particular purpose.: type: select options: - Yes - No You accept full responsibility for your use of the datasets and shall defend and indemnify Carnegie Mellon University, including its employees, officers and agents, against any and all claims arising from your use of the datasets, including but not limited to your use of any copyrighted videos or images that you may create from the datasets.: type: select options: - Yes - No You will treat people appearing in this data with respect and dignity.: type: select options: - Yes - No This data comes with no warranty or guarantee of any kind, and you accept full liability.: type: select options: - Yes - No extra_gated_heading: "TAO-Amodal VIDEO Request" extra_gated_button_content: "Request Data" --- # TAO-Amodal Dataset Official Source for Downloading the TAO-Amodal Dataset. [**πŸ“™ Project Page**](https://tao-amodal.github.io/) | [**πŸ’» Code**](https://github.com/WesleyHsieh0806/TAO-Amodal) | [**πŸ“Ž Paper Link**](https://arxiv.org/abs/2312.12433) | [**✏️ Citations**](#citations)
TAO-Amodal

Contact: [πŸ™‹πŸ»β€β™‚οΈCheng-Yen (Wesley) Hsieh](https://wesleyhsieh0806.github.io/) ## Dataset Description Our dataset augments the TAO dataset with amodal bounding box annotations for fully invisible, out-of-frame, and occluded objects. Note that this implies TAO-Amodal also includes modal segmentation masks (as visualized in the color overlays above). Our dataset encompasses 880 categories, aimed at assessing the occlusion reasoning capabilities of current trackers through the paradigm of Tracking Any Object with Amodal perception (TAO-Amodal). ### Dataset Download 1. Download with git: ```bash git lfs install git clone git@hf.co:datasets/chengyenhsieh/TAO-Amodal ``` - Download with [`python`](https://huggingface.co/docs/huggingface_hub/guides/download#download-files-from-the-hub): ```python from huggingface_hub import snapshot_download snapshot_download(repo_id="chengyenhsieh/TAO-Amodal") ``` 2. Unzip all videos: Modify `dataset_root` in [unzip_video.py](./unzip_video.py) and run: ```bash python unzip_video.py ``` 3. Download video frames through git (optional): You can either download the frames following the instructions [here](https://motchallenge.net/tao_download.php) (recommended) or modify our provided [script](./download_frames.sh) and run ```bash bash download_frames.sh ``` ## πŸ“š Dataset Structure The dataset should be structured like this: ```bash TAO-Amodal β”œβ”€β”€ frames β”‚ └── train β”‚ β”œβ”€β”€ ArgoVerse β”‚ β”œβ”€β”€ BDD β”‚ β”œβ”€β”€ Charades β”‚ β”œβ”€β”€ HACS β”‚ β”œβ”€β”€ LaSOT β”‚ └── YFCC100M β”œβ”€β”€ amodal_annotations β”‚ β”œβ”€β”€ train/validation/test.json β”‚ β”œβ”€β”€ train_lvis_v1.json β”‚ └── validation_lvis_v1.json β”œβ”€β”€ example_output β”‚ └── prediction.json β”œβ”€β”€ BURST_annotations β”‚ β”œβ”€β”€ train β”‚ └── train_visibility.json β”‚ ... ``` ## πŸ“š File Descriptions | File Name | Description | | -------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | train/validation/test.json | Formal annotation files. We use these annotations for visualization. Categories include those in [lvis](https://www.lvisdataset.org/) v0.5 and freeform categories. | | train_lvis_v1.json | We use this file to train our [amodal-expander](https://tao-amodal.github.io/index.html#Amodal-Expander), treating each image frame as an independent sequence. Categories are aligned with those in lvis v1.0. | | validation_lvis_v1.json | We use this file to evaluate our [amodal-expander](https://tao-amodal.github.io/index.html#Amodal-Expander). Categories are aligned with those in lvis v1.0. | | prediction.json | Example output json from amodal-expander. Tracker predictions should be structured like this file to be evaluated with our [evaluation toolkit](https://github.com/WesleyHsieh0806/TAO-Amodal?tab=readme-ov-file#bar_chart-evaluation). | | BURST_annotations/XXX.json | Modal mask annotations from [BURST dataset](https://github.com/Ali2500/BURST-benchmark) with our heuristic visibility attributes. We provide these files for the convenience of visualization | ### Annotation and Prediction Format Our annotations are structured similarly as [TAO](https://github.com/TAO-Dataset/tao/blob/master/tao/toolkit/tao/tao.py#L4) with some modifications. Annotations: ```bash Annotation file format: { "info" : info, "images" : [image], "videos": [video], "tracks": [track], "annotations" : [annotation], "categories": [category], "licenses" : [license], } annotation: { "id": int, "image_id": int, "track_id": int, "bbox": [x,y,width,height], "area": float, # Redundant field for compatibility with COCO scripts "category_id": int, "video_id": int, # Other important attributes for evaluation on TAO-Amodal "amodal_bbox": [x,y,width,height], "amodal_is_uncertain": bool, "visibility": float, (0.~1.0) } image, info, video, track, category, licenses, : Same as TAO ``` Predictions should be structured as: ```bash [{ "image_id" : int, "category_id" : int, "bbox" : [x,y,width,height], "score" : float, "track_id": int, "video_id": int }] ``` Refer to the instructions of [TAO dataset](https://github.com/TAO-Dataset/tao/blob/master/docs/evaluation.md) for further details ## πŸ“Ί Example Sequences Check [here](https://tao-amodal.github.io/#TAO-Amodal) for more examples and [here](https://github.com/WesleyHsieh0806/TAO-Amodal?tab=readme-ov-file#artist-visualization) for visualization code. [](https://tao-amodal.github.io/dataset.html "tao-amodal") ## Citation ``` @misc{hsieh2023tracking, title={Tracking Any Object Amodally}, author={Cheng-Yen Hsieh and Tarasha Khurana and Achal Dave and Deva Ramanan}, year={2023}, eprint={2312.12433}, archivePrefix={arXiv}, primaryClass={cs.CV} } ```
Please also cite TAO and BURST dataset if you use our dataset ``` @inproceedings{dave2020tao, title={Tao: A large-scale benchmark for tracking any object}, author={Dave, Achal and Khurana, Tarasha and Tokmakov, Pavel and Schmid, Cordelia and Ramanan, Deva}, booktitle={Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part V 16}, pages={436--454}, year={2020}, organization={Springer} } @inproceedings{athar2023burst, title={Burst: A benchmark for unifying object recognition, segmentation and tracking in video}, author={Athar, Ali and Luiten, Jonathon and Voigtlaender, Paul and Khurana, Tarasha and Dave, Achal and Leibe, Bastian and Ramanan, Deva}, booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision}, pages={1674--1683}, year={2023} } ```