search

FishMaster93 / U-FFIA

The audio-visual fusion method for FFIA
8 stars 0 forks source link

Data Directory #3

Open chandlerbing65nm opened 1 month ago

chandlerbing65nm commented 1 month ago

Hi, I downloaded your dataset. Can you give insight on how the data are structured and their naming?

When I printed the folder tree, this is what I've seen:

2022_6_13/
│
├── PM_70/
│   └── none/
│       ├── 13_audio_2.wav
│       ├── 13_audio_22.wav
│       └── ...
│
└── AM_40/
    └── none/
        ├── 13_audio_2.wav
        ├── 13_audio_22.wav
        └── ...

2022_6_15/
│
├── PM_100/
│   ├── none/
│   │   ├── 15_audio_18.wav
│   │   ├── 15_audio_19.wav
│   │   └── ...
│   ├── weak/
│   │   ├── 15_audio_18.wav
│   │   ├── 15_audio_19.wav
│   │   └── ...
│   ├── medium/
│   │   ├── 15_audio_81.wav
│   │   ├── 15_audio_87.wav
│   │   └── ...
│   └── strong/
│       ├── 15_audio_81.wav
│       ├── 15_audio_87.wav
│       └── ...
│
└── AM_40/
    ├── none/
    │   ├── 15_audio_18.wav
    │   ├── 15_audio_19.wav
    │   └── ...
    ├── weak/
    │   ├── 15_audio_18.wav
    │   ├── 15_audio_19.wav
    │   └── ...
    ├── medium/
    │   ├── 15_audio_18.wav
    │   ├── 15_audio_19.wav
    │   └── ...
    └── strong/
        ├── 15_audio_18.wav
        ├── 15_audio_19.wav
        └── ...

2022_6_16/
│
├── PM_100/
│   ├── none/
│   │   ├── 16_audio_26.wav
│   │   ├── 16_audio_14.wav
│   │   └── ...
│   ├── weak/
│   │   ├── 16_audio_26.wav
│   │   ├── 16_audio_69.wav
│   │   └── ...
│   ├── medium/
│   │   ├── 16_audio_26.wav
│   │   ├── 16_audio_69.wav
│   │   └── ...
│   └── strong/
│       ├── 16_audio_26.wav
│       ├── 16_audio_43.wav
│       └── ...
│
└── AM_40/
    ├── none/
    │   ├── 16_audio_26.wav
    │   ├── 16_audio_14.wav
    │   └── ...
    ├── weak/
    │   ├── 16_audio_26.wav
    │   ├── 16_audio_14.wav
    │   └── ...
    ├── medium/
    │   ├── 16_audio_26.wav
    │   ├── 16_audio_43.wav
    │   └── ...
    └── strong/
        ├── 16_audio_26.wav
        ├── 16_audio_14.wav
        └── ...

2022_6_17/
│
├── PM_100/
│   ├── none/
│   │   ├── 17_audio_4.wav
│   │   ├── 17_audio_3.wav
│   │   └── ...
│   ├── weak/
│   │   ├── 17_audio_4.wav
│   │   ├── 17_audio_45.wav
│   │   └── ...
│   ├── medium/
│   │   ├── 17_audio_4.wav
│   │   ├── 17_audio_45.wav
│   │   └── ...
│   └── strong/
│       ├── 17_audio_4.wav
│       ├── 17_audio_45.wav
│       └── ...
│
└── AM_40/
    ├── none/
    │   ├── 17_audio_4.wav
    │   ├── 17_audio_31.wav
    │   └── ...
    ├── weak/
    │   ├── 17_audio_4.wav
    │   ├── 17_audio_45.wav
    │   └── ...
    ├── medium/
    │   ├── 17_audio_4.wav
    │   ├── 17_audio_45.wav
    │   └── ...
    └── strong/
        ├── 17_audio_4.wav
        ├── 17_audio_45.wav
        └── ...

2022_6_18/
│
└── AM_100/
    └── medium/
        ├── 18_audio_30.wav
        ├── 18_audio_9.wav
        └── ...
chandlerbing65nm commented 1 month ago

For example, in this folder /audio_dataset/2022_6_13/AM_40/none, what does AM_40mean? Additionally, the files inside range from 13_audio_1.wav to 13_audio_30.wav. Does this mean that the data from files 1 to 30 are sequentially connected without gaps?

FishMaster93 commented 1 month ago

Thank you for your interest in our research. In our dataset naming convention, "AM_40" indicates that the data was collected in the morning with 40 fish in the tank. We conducted experiments using various fish populations (15, 40, 70, and 100) in similar tanks to investigate how group size affects feeding behavior. 13_audio_1.wav to 13_audio_30.wav means we cut the continuous feeding video or audio without gaps. 1-30 represents the cutting video or audio id.

chandlerbing65nm commented 1 month ago

@FishMaster93 Thank you for the prompt response. Regarding the folders, when I unzipped the audio folder, the contents of the 2022_6_18folder included only the AM_100folder, which contains only the medium class. For the 2022_6_13folder, across both the AM and PM subfolders, there is only the none class inside them. The 2022_6_15, 2022_6_16, and 2022_6_17 folders contain complete class subfolders. Is this correct?

FishMaster93 commented 1 month ago

Hi, the 2022_6_13 only contains none, because this is the first collected, the fish did not adapt to the new environment. 2022_6_18 contains all datasets, we have added this folder to the original link. The 2022_6_15, 2022_6_16, and 2022_6_17 folders contain complete class subfolders.

chandlerbing65nm commented 1 month ago

2022_6_18 contains all datasets

You mean all the class subfolders? I only see the medium class when I unzipped the file.

FishMaster93 commented 1 month ago

Yes, contains all class subfolders, I don't know whether some problems when you unzipped the file. But I upload a new dataset on 2022_6_18, you can download it: https://zenodo.org/records/11059975. In addition, all audio datasets are corresponding video datasets, you can also compare them with video files