updated readme
Browse files
README.md
CHANGED
@@ -51,14 +51,14 @@ tags:
|
|
51 |
## Dataset Description
|
52 |
|
53 |
- **Homepage:** [Renumics Homepage](https://renumics.com/)
|
54 |
-
- **
|
|
|
55 |
- **Paper** [MIMII DG](https://arxiv.org/abs/2205.13879)
|
56 |
- **Paper** [ToyADMOS2](https://arxiv.org/abs/2106.02369)
|
|
|
57 |
|
58 |
### Dataset Summary
|
59 |
|
60 |
-
[//]: # (todo)
|
61 |
-
|
62 |
[//]: # (todo: verantwortlichkeit)
|
63 |
|
64 |
### Explore the data with Spotlight
|
@@ -78,16 +78,18 @@ df = ds.to_pandas()
|
|
78 |
spotlight.show(df, dtype={'path': spotlight.Audio})
|
79 |
```
|
80 |
|
81 |
-
advanced view
|
82 |
```jupyterpython
|
|
|
83 |
from datasets import load_dataset, load_dataset_builder
|
84 |
from renumics import spotlight
|
85 |
|
86 |
-
|
|
|
87 |
db = load_dataset_builder("renumics/dcase23-task2-enriched", "dev")
|
88 |
|
89 |
-
df = db.config.to_spotlight(
|
90 |
-
spotlight.show(df, dtype={'audio': spotlight.Audio
|
91 |
```
|
92 |
|
93 |
[//]: # (todo: add embeddings column)
|
@@ -99,7 +101,6 @@ spotlight.show(df, dtype={'audio': spotlight.Audio, 'baseline-embeddings': spotl
|
|
99 |
For each instance, there is a Audio for the audio, a string for the path, an integer for the section, a string for the d1p (parameter), a string for the d1v (value),
|
100 |
a ClassLabel for the label and a ClassLabel for the class.
|
101 |
|
102 |
-
[//]: # (todo)
|
103 |
```
|
104 |
{'audio': {'array': array([ 0. , 0.00024414, -0.00024414, ..., -0.00024414,
|
105 |
0. , 0. ], dtype=float32),
|
@@ -110,12 +111,13 @@ a ClassLabel for the label and a ClassLabel for the class.
|
|
110 |
'section': 1
|
111 |
'd1p': 'f-n'
|
112 |
'd1v': 'A'
|
113 |
-
'd2p': ''
|
114 |
-
'd2v': ''
|
115 |
-
'd3p': ''
|
116 |
-
'd3v': ''
|
|
|
117 |
'label': 0 (normal)
|
118 |
-
'class':
|
119 |
}
|
120 |
```
|
121 |
|
@@ -130,19 +132,18 @@ The length of each audio file is 10 seconds.
|
|
130 |
- `section`: an integer
|
131 |
- `d*p`:
|
132 |
- `d*v`:
|
|
|
133 |
- `class`:
|
134 |
- `label`: an integer whose value may be either _0_, indicating that the audio sample is _normal_, _1_, indicating that the audio sample contains an _anomaly_.
|
135 |
|
136 |
-
[//]: # (todo)
|
137 |
-
|
138 |
### Data Splits
|
139 |
|
140 |
The development dataset has 2 splits: _train_ and _test_.
|
141 |
|
142 |
-
| Dataset Split | Number of Instances in Split |
|
143 |
-
| -------------
|
144 |
-
| Train |
|
145 |
-
| Test |
|
146 |
|
147 |
The information for the evaluation dataset will follow after release.
|
148 |
|
@@ -221,19 +222,17 @@ Not applicable.
|
|
221 |
|
222 |
### Baseline system
|
223 |
|
224 |
-
[
|
225 |
-
Two baseline systems are available on the Github repository. The baseline systems provide a simple entry-level approach that gives a reasonable performance in the dataset of Task 2. They are good starting points, especially for entry-level researchers who want to get familiar with the anomalous-sound-detection task.
|
226 |
|
227 |
### Dataset Curators
|
228 |
|
229 |
[//]: # (todo)
|
230 |
Example: The SNLI corpus was developed by Samuel R. Bowman, Gabor Angeli, Christopher Potts, and Christopher D. Manning as part of the [Stanford NLP group](https://nlp.stanford.edu/).
|
231 |
|
232 |
-
It was supported by a Google Faculty Research Award, a gift from Bloomberg L.P., the Defense Advanced Research Projects Agency (DARPA) Deep Exploration and Filtering of Text (DEFT) Program under Air Force Research Laboratory (AFRL) contract no. FA8750-13-2-0040, the National Science Foundation under grant no. IIS 1159679, and the Department of the Navy, Office of Naval Research, under grant no. N00014-10-1-0109.
|
233 |
-
|
234 |
### Licensing Information - Condition of use
|
235 |
|
236 |
-
|
|
|
237 |
|
238 |
|
239 |
### Citation Information (original)
|
@@ -242,6 +241,7 @@ If you use this dataset, please cite all the following papers. We will publish a
|
|
242 |
|
243 |
- Kota Dohi, Tomoya Nishida, Harsh Purohit, Ryo Tanabe, Takashi Endo, Masaaki Yamamoto, Yuki Nikaido, and Yohei Kawaguchi. MIMII DG: sound dataset for malfunctioning industrial machine investigation and inspection for domain generalization task. In arXiv e-prints: 2205.13879, 2022. [[URL](https://arxiv.org/abs/2205.13879)]
|
244 |
- Noboru Harada, Daisuke Niizumi, Daiki Takeuchi, Yasunori Ohishi, Masahiro Yasuda, and Shoichiro Saito. ToyADMOS2: another dataset of miniature-machine operating sounds for anomalous sound detection under domain shift conditions. In Proceedings of the 6th Detection and Classification of Acoustic Scenes and Events 2021 Workshop (DCASE2021), 1–5. Barcelona, Spain, November 2021. [[URL](https://dcase.community/documents/workshop2021/proceedings/DCASE2021Workshop_Harada_6.pdf)]
|
|
|
245 |
|
246 |
```
|
247 |
@dataset{kota_dohi_2023_7687464,
|
@@ -262,4 +262,4 @@ If you use this dataset, please cite all the following papers. We will publish a
|
|
262 |
doi = {10.5281/zenodo.7687464},
|
263 |
url = {https://doi.org/10.5281/zenodo.7687464}
|
264 |
}
|
265 |
-
```
|
|
|
51 |
## Dataset Description
|
52 |
|
53 |
- **Homepage:** [Renumics Homepage](https://renumics.com/)
|
54 |
+
- **Homepage** [DCASE23 Task 2 Challenge](https://dcase.community/challenge2023/task-first-shot-unsupervised-anomalous-sound-detection-for-machine-condition-monitoring#evaluation)
|
55 |
+
- **Original Dataset Upload (Dev)** [ZENODO: DCASE 2023 Challenge Task 2 Development Dataset](https://zenodo.org/record/7687464#.Y_9VtdLMLmE)
|
56 |
- **Paper** [MIMII DG](https://arxiv.org/abs/2205.13879)
|
57 |
- **Paper** [ToyADMOS2](https://arxiv.org/abs/2106.02369)
|
58 |
+
- **Paper** [First-shot anomaly detection for machine condition monitoring: A domain generalization baseline](https://arxiv.org/pdf/2303.00455.pdf)
|
59 |
|
60 |
### Dataset Summary
|
61 |
|
|
|
|
|
62 |
[//]: # (todo: verantwortlichkeit)
|
63 |
|
64 |
### Explore the data with Spotlight
|
|
|
78 |
spotlight.show(df, dtype={'path': spotlight.Audio})
|
79 |
```
|
80 |
|
81 |
+
to load a more advanced view, we can utilize spotlight's layout function and add clear names for the machine_types in the dataset.
|
82 |
```jupyterpython
|
83 |
+
import pandas as pd
|
84 |
from datasets import load_dataset, load_dataset_builder
|
85 |
from renumics import spotlight
|
86 |
|
87 |
+
train = load_dataset("renumics/dcase23-task2-enriched", "dev", split="train", streaming=False)
|
88 |
+
test = load_dataset("renumics/dcase23-task2-enriched", "dev", split="test", streaming=False)
|
89 |
db = load_dataset_builder("renumics/dcase23-task2-enriched", "dev")
|
90 |
|
91 |
+
df = pd.concat([db.config.to_spotlight(train), db.config.to_spotlight(test)])
|
92 |
+
spotlight.show(df, dtype={'audio': spotlight.Audio}, layout=db.config.get_layout())
|
93 |
```
|
94 |
|
95 |
[//]: # (todo: add embeddings column)
|
|
|
101 |
For each instance, there is a Audio for the audio, a string for the path, an integer for the section, a string for the d1p (parameter), a string for the d1v (value),
|
102 |
a ClassLabel for the label and a ClassLabel for the class.
|
103 |
|
|
|
104 |
```
|
105 |
{'audio': {'array': array([ 0. , 0.00024414, -0.00024414, ..., -0.00024414,
|
106 |
0. , 0. ], dtype=float32),
|
|
|
111 |
'section': 1
|
112 |
'd1p': 'f-n'
|
113 |
'd1v': 'A'
|
114 |
+
'd2p': 'nan'
|
115 |
+
'd2v': 'nan'
|
116 |
+
'd3p': 'nan'
|
117 |
+
'd3v': 'nan'
|
118 |
+
'domain': 0 (source)
|
119 |
'label': 0 (normal)
|
120 |
+
'class': 1 (fan)
|
121 |
}
|
122 |
```
|
123 |
|
|
|
132 |
- `section`: an integer
|
133 |
- `d*p`:
|
134 |
- `d*v`:
|
135 |
+
- `domain`:
|
136 |
- `class`:
|
137 |
- `label`: an integer whose value may be either _0_, indicating that the audio sample is _normal_, _1_, indicating that the audio sample contains an _anomaly_.
|
138 |
|
|
|
|
|
139 |
### Data Splits
|
140 |
|
141 |
The development dataset has 2 splits: _train_ and _test_.
|
142 |
|
143 |
+
| Dataset Split | Number of Instances in Split | Source Domain / Target Domain Samples |
|
144 |
+
| ------------- |------------------------------|---------------------------------------|
|
145 |
+
| Train | 7000 | 6930 / 70 |
|
146 |
+
| Test | 1400 | 700 / 700 |
|
147 |
|
148 |
The information for the evaluation dataset will follow after release.
|
149 |
|
|
|
222 |
|
223 |
### Baseline system
|
224 |
|
225 |
+
The baseline system is available on the Github repository [dcase2023_task2_baseline_ae](https://github.com/nttcslab/dase2023_task2_baseline_ae).The baseline systems provide a simple entry-level approach that gives a reasonable performance in the dataset of Task 2. They are good starting points, especially for entry-level researchers who want to get familiar with the anomalous-sound-detection task.
|
|
|
226 |
|
227 |
### Dataset Curators
|
228 |
|
229 |
[//]: # (todo)
|
230 |
Example: The SNLI corpus was developed by Samuel R. Bowman, Gabor Angeli, Christopher Potts, and Christopher D. Manning as part of the [Stanford NLP group](https://nlp.stanford.edu/).
|
231 |
|
|
|
|
|
232 |
### Licensing Information - Condition of use
|
233 |
|
234 |
+
This is a feature/embeddings-enriched version of the "DCASE 2023 Challenge Task 2 Development Dataset".
|
235 |
+
The [original dataset](https://dcase.community/challenge2023/task-first-shot-unsupervised-anomalous-sound-detection-for-machine-condition-monitoring#audio-datasets) was created jointly by **Hitachi, Ltd.** and **NTT Corporation** and is available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license.
|
236 |
|
237 |
|
238 |
### Citation Information (original)
|
|
|
241 |
|
242 |
- Kota Dohi, Tomoya Nishida, Harsh Purohit, Ryo Tanabe, Takashi Endo, Masaaki Yamamoto, Yuki Nikaido, and Yohei Kawaguchi. MIMII DG: sound dataset for malfunctioning industrial machine investigation and inspection for domain generalization task. In arXiv e-prints: 2205.13879, 2022. [[URL](https://arxiv.org/abs/2205.13879)]
|
243 |
- Noboru Harada, Daisuke Niizumi, Daiki Takeuchi, Yasunori Ohishi, Masahiro Yasuda, and Shoichiro Saito. ToyADMOS2: another dataset of miniature-machine operating sounds for anomalous sound detection under domain shift conditions. In Proceedings of the 6th Detection and Classification of Acoustic Scenes and Events 2021 Workshop (DCASE2021), 1–5. Barcelona, Spain, November 2021. [[URL](https://dcase.community/documents/workshop2021/proceedings/DCASE2021Workshop_Harada_6.pdf)]
|
244 |
+
- Noboru Harada, Daisuke Niizumi, Daiki Takeuchi, Yasunori Ohishi, and Masahiro Yasuda. First-shot anomaly detection for machine condition monitoring: a domain generalization baseline. In arXiv e-prints: 2303.00455, 2023. [[URL](https://arxiv.org/abs/2303.00455.pdf)]
|
245 |
|
246 |
```
|
247 |
@dataset{kota_dohi_2023_7687464,
|
|
|
262 |
doi = {10.5281/zenodo.7687464},
|
263 |
url = {https://doi.org/10.5281/zenodo.7687464}
|
264 |
}
|
265 |
+
```
|