feat: Describe pictures using vision models (#259)

* draft for picture description models

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* vlm description using AutoModelForVision2Seq

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* add generation options

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* update vlm API

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* allow only localhost traffic

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* rename model

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* do not run with vlm api

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* more renaming

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* fix examples path

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* apply CLI download login

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* fix name of cli argument

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* use with_smolvlm in models download

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

---------

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

docling/cli/main.py

@@ -226,6 +226,10 @@ def convert(
             help="Enable the picture classification enrichment model in the pipeline.",
         ),
     ] = False,
+    enrich_picture_description: Annotated[
+        bool,
+        typer.Option(..., help="Enable the picture description model in the pipeline."),
+    ] = False,
     artifacts_path: Annotated[
         Optional[Path],
         typer.Option(..., help="If provided, the location of the model artifacts."),
@@ -382,6 +386,7 @@ def convert(
             do_table_structure=True,
             do_code_enrichment=enrich_code,
             do_formula_enrichment=enrich_formula,
+            do_picture_description=enrich_picture_description,
             do_picture_classification=enrich_picture_classes,
             document_timeout=document_timeout,
         )