# OpenSet

**Repository Path**: ctguhzy/OpenSet

## Basic Information

- **Project Name**: OpenSet
- **Description**: OpenSet for dataset preparing.
- **Primary Language**: Python
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-05-10
- **Last Updated**: 2026-05-15

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# OpenSet Video Object Detection Dataset Preparation Tool

## Prepare Source Data

1. Stop Docker Containers
   ```bash
   cd /mnt/sda/users/hzy/MyCode/AIPRS/docker
   docker-compose -f docker-scale.yml down
   ```

2. Start Data Containers
   ```bash
   docker-compose -f docker-scale.yml up -d --scale app=0
   ```

3. Enter the Container Container and Prepare Source Data
   ```bash
   docker exec -it DB bash
   ```
   * Prepare Source Data in DB
   ```bash
   cd /data

   tar -cvzf videos202605xx.tar.gz videos
   ```

   * Remove used data
   ```bash
   rm -rf videos/*
   rm -rf images/*
   ```

   * Remove video table data in Postgres Dataset
   ```bash
   psql -U aiprs -d aiprs -c "TRUNCATE TABLE videos restart identity;"
   ```

   * Exit the Container Container
   ```bash
   exit
   ```

4. Enter OpenSet folder and Copy source data to local machine
   ```bash
   cd OpenSet
   docker cp DB:/data/videos202605xx.tar.gz data/videos202605xx.tar.gz
   ```

5. Extract video data
   ```bash
   tar -xzvf data/videos202605xx.tar.gz -C data
   ```

## Extract video frames from videos
   ```bash
   cd OpenSet
   python scripts/extract_video_frames.py
   ```

## Remove similar video frames
   ```bash
   cd OpenSet
   python scripts/remove_similar.py
   ```

## Automatically label the frame images in data/frames with YOLO model
   ```bash
   cd OpenSet
   python scripts/auto_labeler.py --model AIPRS_XXXX.pt
   ```

## Output labeled data
   ```bash
   cd OpenSet
   python scripts/process_dataset.py
   ```

   Then you have a labeled dataset in output folder. You need to move the dataset to your laptop and Check it manually with DarkLabel.