GTA_dataset

gta_crime_dataset

Number of scenes: 198

Counter of videos: Total: 961

Arrest: 134

Arson : 148

Assault: 107

Fight: 126

Vandalism: 105

Explosion: 158

Robbery: 48

Shooting: 140

S3 bucket:

s3://gta-crime-dataset

Steps:

1. clone repository, install requirements.txt
2. Register on aws, create account
3. configure aws cli: instruction AWS Access Key ID: I will tell you when I will know your account ID AWS Secret Access Key: same
4. Now you can use dvc

Dataset structure

For each class separate folder

├── 1      # scene number
│   ├── Fight1.mp4    # scene variants (angle, duration changes)
│   ├── Fight2.mp4
│   ├── Fight3.mp4
│   └── Fight4\ (2).mp4
├── 2
│   ├── Fight1.mp4
│   ├── Fight2.mp4
│   ├── Fight3.mp4
│   └── Fight4.mp4
├── 3
│   ├── Fight1_2A.mp4
│   ├── Fight2_2A.mp4
│   ├── Fight3_2A.mp4
│   ├── Fight5_2A.mp4
│   ├── Fight6_2A.mp4
│   └── Fight_204_2A.mp4
├── 4
│   ├── Cinema_20fight_202_2A.mp4
│   ├── Cinema_20fight_203_2A.mp4
│   ├── Cinema_20fight_204_2A.mp4
│   ├── Cinema_20fight_205_2A.mp4
│   ├── Cinema_20fight_206_2A.mp4
│   └── Cinema_20fighting_201_2A.mp4
├── 5
│   ├── Shop_20fight_201_2A.mp4
│   ├── Shop_20fight_202_2A.mp4
│   ├── Shop_20fight_203_2A.mp4
│   ├── Shop_20fight_204_2A.mp4
│   ├── Shop_20fight_205_2A.mp4
│   └── Shop_20fight_206_2A.mp4
└── 6
    ├── Fighting1_2A.mp4
    ├── Fighting_202_2A.mp4
    ├── Fighting_203_2A.mp4
    ├── Fighting_204_2A.mp4
    ├── Fighting_205_2A.mp4
    └── Fighting_206_2A.mp4

DVC

Documentation: https://dvc.org

You do not need to do any configuration

How to pull all data

1. dvc pull * -r s3remote

pull only one class

2. dvc pull Fight.dvc

How to add data

1. dvc add filename
2. dvc push -r s3remote
3. make sure that data is really loading to bucket
4. add recently created File.dvc, .gitignore to git
5. push changes

How to update existing data (example )

1. dvc pull * (or particular file)
2. update folder with new videos
3. dvc update Fight.dvc (Class name)
4. dvc push -r s3remote