Open sourcing of source code and results is ongoing.
Download SV248S dataset from the official website [SV248S] The dataset should look like this:
${data_root}
├── 01_000000
├── 01_000001
├── 01_000002
...
├── 06_000045
├── 06_000046
├── 06_000047
└── list.txt
modify the root
to the ${data_root}
in demo_fast.py
The list.txt
file contains sequence names of the datasets.
Simplely create it with
ls > list.txt
We provide scripts to link original SV248S dataset to this data structure.
python demo_fast.py
modify the demo_fast.py and set use_time
to False