Downloads · 30 days
12
27% of all-time downloads
trocr-base-handwritten-finetuned
dodo1234/trocr-base-handwritten-finetuned
trocr-base-handwritten-finetuned is a image-to-text model from dodo1234. Use it when you need a caption or text from an image. The card lists the license as mit.
TrOCR fine-tuné pour la reconnaissance d'écriture manuscrite, à partir de microsoft/trocr-base-stage1.
All-time downloads
45
Public
Parameters
385M
1.5 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors1.5 GB · 100%
At a glance
- Task
- Image-to-Text
- License
- mit
- Model type
- vision-encoder-decoder
- Access
- Public
- Created
- Sep 16, 2026
- Updated
- Sep 16, 2026
- SHA
- 2d44602d
Base models
Model card
From the Hugging Face model README
trocr-base-handwritten-finetuned
TrOCR fine-tuné pour la reconnaissance d'écriture manuscrite, à partir de microsoft/trocr-base-stage1.
Résultats sur le jeu de test (778 exemples)
| Métrique | Valeur |
|---|---|
| CER | 50.15 % |
| WER | 94.96 % |
| Exact match | 0.00 % |
Utilisation
from transformers import VisionEncoderDecoderModel, TrOCRProcessor
from PIL import Image
processor = TrOCRProcessor.from_pretrained("dodo1234/trocr-base-handwritten-finetuned")
model = VisionEncoderDecoderModel.from_pretrained("dodo1234/trocr-base-handwritten-finetuned")
image = Image.open("ligne_manuscrite.jpg").convert("RGB")
pixel_values = processor(image, return_tensors="pt").pixel_values
ids = model.generate(pixel_values)
print(processor.batch_decode(ids, skip_special_tokens=True)[0])
Entraînement
- Base : microsoft/trocr-base-stage1
- Batch size : 8, fp16, Seq2SeqTrainer
- Task
- Image-to-Text
- Type
- vision-encoder-decoder
- License
- mit
- Languages
- fr
- Created
- Sep 16, 2026
- Updated
- Sep 16, 2026