Update README.md
Browse files
README.md
CHANGED
|
@@ -6,30 +6,37 @@ tags:
|
|
| 6 |
- bird classification
|
| 7 |
- passive acoustic monitoring
|
| 8 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
We present the GADME benchmark that covers a comprehensive range of avian monitoring datasets.
|
| 10 |
We offer a static set of evaluation datasets and a varied collection of training datasets, enabling the application of diverse methodologies
|
| 11 |
|
| 12 |
### Datasets
|
| 13 |
|
| 14 |
-
#### Train
|
| 15 |
- Exclusively using focal audio data from Xeno-Canto (XC) with quality ratings A, B, C and excluding all recordings that are CC-ND.
|
| 16 |
- Each dataset is tailored for specific target species identified in soundscape files.
|
| 17 |
- We offer detected events and corresponding cluster assignments to identify bird sounds in each recording.
|
| 18 |
- We provide the full recordings from XC! These can generate multiple samples from a single instance.
|
| 19 |
|
| 20 |
-
#### Test
|
| 21 |
- Only soundscape data sourced from Zenodo.
|
| 22 |
- We provide the full recording with the complete label set and specified bounding boxes.
|
| 23 |
- This dataset excludes recordings that do not contain bird calls ("no_call").
|
| 24 |
- Task: Multiclass ("ebird_code")
|
| 25 |
|
| 26 |
-
#### Test_5s
|
| 27 |
- Only soundscape data from Zenodo formatted acoording to the Kaggle evaluation scheme.
|
| 28 |
- Each recording is segmented into 5-second intervals without overlaps.
|
| 29 |
- This contains the "no_call" class.
|
| 30 |
- Task: Multilabel ("ebird_code_multilabel")
|
| 31 |
|
| 32 |
-
### Subsets
|
| 33 |
|
| 34 |
Numbers need to be updated
|
| 35 |
|
|
|
|
| 6 |
- bird classification
|
| 7 |
- passive acoustic monitoring
|
| 8 |
---
|
| 9 |
+
## Dataset Description
|
| 10 |
+
|
| 11 |
+
- **Repository:** [https://github.com/s3prl/s3prl](https://github.com/s3prl/s3prl)
|
| 12 |
+
- **Paper:** [GADME](https://arxiv.org/))
|
| 13 |
+
- **Point of Contact:** [Lukas Rauch](mailto:lukas.rauch@uni-kassel.de) and [Moritz Wirth](mailto:moritz.wirth@uni-kassel.de)
|
| 14 |
+
|
| 15 |
+
### Dataset Summary
|
| 16 |
We present the GADME benchmark that covers a comprehensive range of avian monitoring datasets.
|
| 17 |
We offer a static set of evaluation datasets and a varied collection of training datasets, enabling the application of diverse methodologies
|
| 18 |
|
| 19 |
### Datasets
|
| 20 |
|
| 21 |
+
##### Train
|
| 22 |
- Exclusively using focal audio data from Xeno-Canto (XC) with quality ratings A, B, C and excluding all recordings that are CC-ND.
|
| 23 |
- Each dataset is tailored for specific target species identified in soundscape files.
|
| 24 |
- We offer detected events and corresponding cluster assignments to identify bird sounds in each recording.
|
| 25 |
- We provide the full recordings from XC! These can generate multiple samples from a single instance.
|
| 26 |
|
| 27 |
+
##### Test
|
| 28 |
- Only soundscape data sourced from Zenodo.
|
| 29 |
- We provide the full recording with the complete label set and specified bounding boxes.
|
| 30 |
- This dataset excludes recordings that do not contain bird calls ("no_call").
|
| 31 |
- Task: Multiclass ("ebird_code")
|
| 32 |
|
| 33 |
+
##### Test_5s
|
| 34 |
- Only soundscape data from Zenodo formatted acoording to the Kaggle evaluation scheme.
|
| 35 |
- Each recording is segmented into 5-second intervals without overlaps.
|
| 36 |
- This contains the "no_call" class.
|
| 37 |
- Task: Multilabel ("ebird_code_multilabel")
|
| 38 |
|
| 39 |
+
#### Subsets
|
| 40 |
|
| 41 |
Numbers need to be updated
|
| 42 |
|