Evaluate DeepOCR - Paddle OCR - (Data Scientist or ML engineer)
Budget: €36 – €0 EUR
Goal:
Paddle OCR is an open source DeepOCR
We would like to find out if it performs well on European characters and if it fits into our current architecture
As a first proposal we see following steps needed:
Align approach with team - most likely:
Get local copy
Install Paddle OCR
Create Test Set for benchmark (European characters, semi-structured docs)
Present benchmark to the team
Try to train Paddle OCR for a found weakness and see if it learns accordingly
Summarize your findings and write a report
OPTIONAL:
If OK -> integrate Paddle OCR into our architecture (separate service and impl. of given interface in C#)
Re-run our service with Paddle OCR with a given set of benchmarks
Paddle OCR is an open source DeepOCR
We would like to find out if it performs well on European characters and if it fits into our current architecture
As a first proposal we see following steps needed:
Align approach with team - most likely:
Get local copy
Install Paddle OCR
Create Test Set for benchmark (European characters, semi-structured docs)
Present benchmark to the team
Try to train Paddle OCR for a found weakness and see if it learns accordingly
Summarize your findings and write a report
OPTIONAL:
If OK -> integrate Paddle OCR into our architecture (separate service and impl. of given interface in C#)
Re-run our service with Paddle OCR with a given set of benchmarks