💡 [REQUEST] - Add better support for Brazilian Portuguese

[insinfo]

起始日期 | Start Date

No response

实现PR | Implementation PR

No response

相关Issues | Reference Issues

No response

摘要 | Summary

I did a test to OCR scanned documents in Brazilian Portuguese, and I saw that MiniCPM-V makes a lot of mistakes on scanned documents in Portuguese

基本示例 | Basic Example

https://huggingface.co/spaces/openbmb/MiniCPM-V-2_6

Prompt: transcribe image

result

99803
DO ESTADO RIO DE JANEIRO
Rio das Prefeitura Municipal de Ostras
PROTOCOLO GERAL
CENTRO
Processo: Data: Hora: 03/09/2003 10:53:56
Requerente: COSCARELLI E CIA LTDA ME
Sec.Destino: Secretaria Municipal de Fazenda
Dept.Destino: Dept® De Tributos & Fiscalizacao
Assunto: ALVARA

the correct thing would be

Here is the transcription of the text from the image:

ESTADO DO RIO DE JANEIRO
Prefeitura Municipal de Rio das Ostras
PROTOCOLO GERAL

Processo: 18457 / 2003
Data: 03/09/2003
Hora: 10:53:56
Requerente: COSCARELLI E CIA LTDA ME
Sec. Destino: Secretaria Municipal de Fazenda
Dept. Destino: Depto. de Tributos e Fiscalização
Assunto: ALVARÁ

result

ESTADO DO RIO DE JANEIRO
Ostrs Municipal Rio das Prefeitura
PROTOCOLO GERAL
16:18:28 Data 07/2003 Processo
Colonia Pescadores
Cestino S.Pub.
Dept.Destino.C
Assunto, AGRADECIMENTO.FA

the correct thing would be

Here is the transcription of the text from the image:

ESTADO DO RIO DE JANEIRO
Prefeitura Municipal de Rio das Ostras
PROTOCOLO GERAL

Processo: 15314 / 2003
Data: 25/07/2003
Hora: 16:18:28

Requerente: COLÔNIA DE PESCADORES Z-22
Sec. Destino: Sec. Mun. Urbanismo Obras e S. Pub.
Dept. Destino: 0
Assunto: AGRADECIMENTO / FAZ

缺陷 | Drawbacks

more training time and more datasets

未解决问题 | Unresolved questions

No response

[insinfo]

perhaps training on this dataset could help improve accuracy https://zenodo.org/records/7872951