benediktkol commited on
Commit
a6d9ec8
β€’
1 Parent(s): 25ad797

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +92 -17
README.md CHANGED
@@ -1,38 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  # DDOS: The Drone Depth and Obstacle Segmentation Dataset
2
 
3
- ## Overview
4
 
5
- The Drone Depth and Obstacle Segmentation (DDOS) dataset is a collection of aerial images captured by drones, with corresponding depth maps, and annotated with pixel-wise segmentation masks. DDOS is designed to facilitate research and development in the field of computer vision, specifically addressing tasks related to depth estimation and obstacle segmentation from aerial imagery.
6
 
7
- For a detailed understanding of DDOS, please refer to the associated research paper: [DDOS: The Drone Depth and Obstacle Segmentation Dataset](https://arxiv.org/abs/2312.12494).
8
 
9
- **Please Note:** The final documentation for this dataset has not been finalized yet. Therefore, the information provided here is subject to change. This README serves as a preliminary guide to give you an understanding of the dataset's structure and contents.
10
 
11
  ## Data Structure
12
 
13
  DDOS is organised as follows:
14
 
15
  - Data Splits:
16
- - Train: Contains training data for model training.
17
- - Validation: Contains data for validation during model development.
18
- - Test: Contains data for the final evaluation of the trained model.
19
 
20
  - Environments:
21
- - Neighbourhood: Contains data captured in urban and residential environments.
22
  - Park: Contains data captured in park and natural environments.
23
 
24
  - Flights:
25
  - Each flight is represented by a unique flight ID and is contained within the corresponding environment directory.
26
 
27
  - Data for Each Flight:
28
- - Depth: Contains depth maps representing the distance of objects from the camera.
29
- - Scene Flow: Contains scene flow data representing the 3D motion of objects.
30
  - Image: Contains RGB images captured by the drone camera.
31
- - Segmentation: Contains pixel-wise segmentation masks for obstacle segmentation.
 
 
32
  - Surface Normal: Contains surface normal maps representing the orientation of object surfaces.
33
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
34
 
35
 
 
36
  ## Additional Information
37
 
38
  **Class Mapping:** The segmentation masks use the following class labels for obstacle segmentation:
@@ -45,25 +116,29 @@ DDOS is organised as follows:
45
  'trees': 180,
46
  'buildings': 160,
47
  'vehicles': 140,
48
- 'people': 120,
49
  'animals': 100,
50
  'other': 80
51
  }
52
  ```
53
- **Metadata:** The dataset contains metadata, such as GPS coordinates, altitude, and camera parameters, which provide valuable contextual information about each flight.
54
 
55
 
 
56
  ## Dataset Usage
57
 
58
- - **Data Loading**
59
- To load and utilize the DDOS dataset for your machine learning projects, you can refer to the official PyTorch data loading tutorial for beginners: [PyTorch Data Loading Tutorial](https://pytorch.org/tutorials/beginner/basics/data_tutorial.html)
60
  This tutorial will guide you through the process of loading data, creating data loaders, and preparing the dataset for training or evaluation using PyTorch.
61
 
62
- - **Disclaimer:** While every effort has been made to ensure the accuracy and quality of the DDOS dataset, the dataset owner and contributors cannot be held responsible for any inaccuracies or errors in the data. Users are advised to exercise caution and due diligence when using the dataset for their research or applications.
 
63
 
 
64
  ## License
65
  DDOS is openly licensed under [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/).
66
 
 
 
67
  ## Citation
68
  If you use DDOS in your research or projects, please cite our paper:
69
 
@@ -74,4 +149,4 @@ If you use DDOS in your research or projects, please cite our paper:
74
  journal={arXiv preprint arXiv:2312.12494},
75
  year={2023}
76
  }
77
- ```
 
1
+ ---
2
+ license: cc-by-nc-4.0
3
+ pretty_name: DDOS
4
+ task_categories:
5
+ - image-segmentation
6
+ - depth-estimation
7
+ tags:
8
+ - drones
9
+ - uav
10
+ - aerial
11
+ - vision
12
+ - wires
13
+ - cables
14
+ - outdoor
15
+ - segmentation
16
+ - semantic segmentation
17
+ - depth
18
+ - weather
19
+ - flying
20
+ size_categories:
21
+ - 10K<n<100K
22
+ ---
23
  # DDOS: The Drone Depth and Obstacle Segmentation Dataset
24
 
25
+ The Drone Depth and Obstacle Segmentation (DDOS) dataset comprises synthetic aerial images captured by drones, along with corresponding depth maps and pixel-wise semantic segmentation masks. DDOS is purpose-built to support research and development in computer vision, focusing on tasks such as depth estimation and obstacle segmentation from aerial imagery. Emphasizing the detection of thin structures like wires and effective navigation in diverse weather conditions, DDOS serves as a valuable resource for advancing algorithms in autonomous drone technology.
26
 
 
27
 
28
+ - **Paper:** [DDOS: The Drone Depth and Obstacle Segmentation Dataset](https://arxiv.org/abs/2312.12494)
29
 
30
+ ---------
31
 
32
  ## Data Structure
33
 
34
  DDOS is organised as follows:
35
 
36
  - Data Splits:
37
+ - Train: Contains 300 flights with a total of 30k images for training.
38
+ - Validation: Contains 20 flights with a total of 2k images for validation during model development.
39
+ - Test: Contains 20 flights with a total of 2k images for the final evaluation of the trained model.
40
 
41
  - Environments:
42
+ - Neighborhood: Contains data captured in urban and residential environments.
43
  - Park: Contains data captured in park and natural environments.
44
 
45
  - Flights:
46
  - Each flight is represented by a unique flight ID and is contained within the corresponding environment directory.
47
 
48
  - Data for Each Flight:
 
 
49
  - Image: Contains RGB images captured by the drone camera.
50
+ - Depth: Contains depth maps representing the distance of objects from the camera. These maps are saved as uint16 PNG images, where pixel values range from 0 to 65535, representing distances from 0 to 100 meters linearly.
51
+ - Segmentation: Contains pixel-wise segmentation masks for semantic segmentation. Classes, as well as their corresponding mappings, are mentioned below.
52
+ - Flow: Contains optical flow data representing the apparent motion of objects between consecutive frames.
53
  - Surface Normal: Contains surface normal maps representing the orientation of object surfaces.
54
 
55
+ Overview of file structure:
56
+ ```
57
+ data/
58
+ β”œβ”€β”€ train/
59
+ β”‚ β”œβ”€β”€ neighbourhood/
60
+ β”‚ β”‚ β”œβ”€β”€ 0/
61
+ β”‚ β”‚ β”‚ β”œβ”€β”€ depth/
62
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ 0.png
63
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ ...
64
+ β”‚ β”‚ β”‚ β”‚ └── 99.png
65
+ β”‚ β”‚ β”‚ β”œβ”€β”€ flow/
66
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ 0.png
67
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ ...
68
+ β”‚ β”‚ β”‚ β”‚ └── 99.png
69
+ β”‚ β”‚ β”‚ β”œβ”€β”€ image/
70
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ 0.png
71
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ ...
72
+ β”‚ β”‚ β”‚ β”‚ └── 99.png
73
+ β”‚ β”‚ β”‚ β”œβ”€β”€ segmentation/
74
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ 0.png
75
+ β”‚ β”‚ β”‚ β”‚ β”œβ”€β”€ ...
76
+ β”‚ β”‚ β”‚ β”‚ └── 99.png
77
+ β”‚ β”‚ β”‚ └── surfacenormals/
78
+ β”‚ β”‚ β”‚ β”œβ”€β”€ 0.png
79
+ β”‚ β”‚ β”‚ β”œβ”€β”€ ...
80
+ β”‚ β”‚ β”‚ └── 99.png
81
+ β”‚ β”‚ β”œβ”€β”€ ...
82
+ β”‚ β”‚ └── 249/
83
+ β”‚ β”‚ └── ...
84
+ β”‚ └── park/
85
+ β”‚ β”œβ”€β”€ 0/
86
+ β”‚ β”‚ β”œβ”€β”€ depth/
87
+ β”‚ β”‚ β”‚ └── ...
88
+ β”‚ β”‚ β”œβ”€β”€ flow/
89
+ β”‚ β”‚ β”‚ └── ...
90
+ β”‚ β”‚ β”œβ”€β”€ image/
91
+ β”‚ β”‚ β”‚ └── ...
92
+ β”‚ β”‚ β”œβ”€β”€ segmentation/
93
+ β”‚ β”‚ β”‚ └── ...
94
+ β”‚ β”‚ └── surfacenormals/
95
+ β”‚ β”‚ └── ...
96
+ β”‚ β”œβ”€β”€ ...
97
+ β”‚ └── 49/
98
+ β”‚ └── ...
99
+ β”œβ”€β”€ validation/
100
+ β”‚ └── ...
101
+ └── test/
102
+ └── ...
103
+ ```
104
 
105
 
106
+ ---------
107
  ## Additional Information
108
 
109
  **Class Mapping:** The segmentation masks use the following class labels for obstacle segmentation:
 
116
  'trees': 180,
117
  'buildings': 160,
118
  'vehicles': 140,
 
119
  'animals': 100,
120
  'other': 80
121
  }
122
  ```
123
+ **Metadata:** The dataset contains metadata, such as coordinates, pose, acceleration, weather conditions and camera parameters, which provide valuable contextual information about each flight.
124
 
125
 
126
+ ---------
127
  ## Dataset Usage
128
 
129
+ - **Data Loading:**
130
+ To load and utilize the DDOS dataset for your projects, you can refer to the official PyTorch data loading tutorial: [PyTorch Data Loading Tutorial](https://pytorch.org/tutorials/beginner/basics/data_tutorial.html)
131
  This tutorial will guide you through the process of loading data, creating data loaders, and preparing the dataset for training or evaluation using PyTorch.
132
 
133
+ - **Respect the Data Splits:**
134
+ Please ensure that the testing data is not used for validation. Mixing these datasets could lead to inaccurate assessments of model performance. Maintaining separate datasets for testing and validation helps ensure reliable evaluation and accurate reporting of results.
135
 
136
+ ---------
137
  ## License
138
  DDOS is openly licensed under [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/).
139
 
140
+
141
+ ---------
142
  ## Citation
143
  If you use DDOS in your research or projects, please cite our paper:
144
 
 
149
  journal={arXiv preprint arXiv:2312.12494},
150
  year={2023}
151
  }
152
+ ```