Juartaurus/poi_Engineering

PipelinePOIEngineering

Repo Signboard segmentation and Text recognition

Download pretrain model:

ss: signboards segmentation str: scene text recognition ir: information retrieval

All the checkpoints are available here: poi_engineering_checkpoints

For each model, copy the checkpoint to the same directory (ir --> ir; ss --> ss; str --> str)

Run:

Setup enviroments:

bash setup.sh

Extract text:

bash extract.sh

Image parser

curl -F "image=@/home/trungtv/sources/POI_Engineering/outputs/20220520_150925/frames/frames_0.jpg" http://localhost:5050/parse_image