From 79a9bb0128193554352e3ef3e84595b295c263b9 Mon Sep 17 00:00:00 2001
From: Robert Sachunsky <sachunsky@informatik.uni-leipzig.de>
Date: Sat, 18 Jul 2026 00:59:57 +0200
Subject: [PATCH] cnn-rnn-ocr: increase default batch size and VRAM limit

---
 src/eynollah/eynollah_ocr.py        | 2 +-
 src/eynollah/model_zoo/model_zoo.py | 2 +-
 src/eynollah/predictor.py           | 6 +++---
 3 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/src/eynollah/eynollah_ocr.py b/src/eynollah/eynollah_ocr.py
index fe0bb1f..748d6a2 100644
--- a/src/eynollah/eynollah_ocr.py
+++ b/src/eynollah/eynollah_ocr.py
@@ -83,7 +83,7 @@ class Eynollah_ocr(Eynollah):
         self.logger = logger if logger else logging.getLogger('eynollah.ocr')
         
         self.min_conf_value_of_textline_text = min_conf_value_of_textline_text
-        self.b_s = batch_size or 2 if tr_ocr else 8
+        self.b_s = batch_size or (2 if tr_ocr else 64)
 
         self.model_zoo = model_zoo
         self.setup_models(device=device)
diff --git a/src/eynollah/model_zoo/model_zoo.py b/src/eynollah/model_zoo/model_zoo.py
index 5a0a867..f230a31 100644
--- a/src/eynollah/model_zoo/model_zoo.py
+++ b/src/eynollah/model_zoo/model_zoo.py
@@ -24,7 +24,7 @@ MODEL_VRAM_LIMITS = {
     "region_fl_np": 1756,
     "table": 1818,
     "reading_order": 632,
-    "ocr": 850,
+    "ocr": 2400, # 850 for bs 8
 }
 
 class EynollahModelZoo:
diff --git a/src/eynollah/predictor.py b/src/eynollah/predictor.py
index 2d892c7..0641e38 100644
--- a/src/eynollah/predictor.py
+++ b/src/eynollah/predictor.py
@@ -1,5 +1,5 @@
 from contextlib import ExitStack
-from typing import List, Dict, Tuple, Union
+from typing import List, Dict, Sequence, Tuple, Union
 import logging
 import logging.handlers
 import multiprocessing as mp
@@ -41,10 +41,10 @@ class Predictor(mp.context.SpawnProcess):
     def input_shape(self):
         return self({})
 
-    def predict(self, data: ArrayT, verbose=0) -> ArrayT:
+    def predict(self, data: Union[Sequence[ArrayT], ArrayT], verbose=0) -> Union[Sequence[ArrayT], ArrayT]:
         return self(data)
 
-    def __call__(self, data: Union[ArrayT, Dict]) -> Union[ArrayT, Tuple]:
+    def __call__(self, data: Union[Sequence[ArrayT], ArrayT, Dict]) -> Union[ArrayT, Tuple]:
         # unusable as per python/cpython#79967
         #with self.jobid.get_lock():
         # would work, but not public: