Skip to content

Latest commit

 

History

History
30 lines (25 loc) · 890 Bytes

README.md

File metadata and controls

30 lines (25 loc) · 890 Bytes

ContainerNumber-OCR

tf

Detection and Recognition Container number.
Detection: Pixel_link + MobileNet_v2
Recognition: LSTM
Speed: RTX 2080Ti 200ms/image(1920*1080)

Sample

Demo

  1. Download pretrained model from Google Drive
  2. Install requirements
pip install -r requirements.txt
  1. run
python containernumber_test_ckpt.py

Dataset

Google Drive

Convert to Openvino Model

python tools/export.py \
  --checkpoint ckpt/recognition_v/model_all.ckpt-146000 \
  --output_dir recognition_v_ov