update

app.py

@@ -40,6 +40,8 @@ def load_predictor(
     reco_arch: str,
     assume_straight_pages: bool,
     straighten_pages: bool,
+    detect_language: bool,
+    load_in_8_bit: bool,
     bin_thresh: float,
     box_thresh: float,
 ) -> OCRPredictor:
@@ -51,6 +53,8 @@ def load_predictor(
         reco_arch: recognition architecture
         assume_straight_pages: whether to assume straight pages or not
         straighten_pages: whether to straighten rotated pages or not
+        detect_language: whether to detect the language of the text
+        load_in_8_bit: whether to load the image in 8 bit mode
         bin_thresh: binarization threshold for the segmentation map
         box_thresh: minimal objectness score to consider a box
 
@@ -63,6 +67,8 @@ def load_predictor(
         reco_arch,
         assume_straight_pages=assume_straight_pages,
         straighten_pages=straighten_pages,
+        detect_language=detect_language,
+        load_in_8_bit=load_in_8_bit,
         export_as_straight_boxes=straighten_pages,
         detect_orientation=not assume_straight_pages,
     )
@@ -117,6 +123,8 @@ def analyze_page(
     reco_arch: str,
     assume_straight_pages: bool,
     straighten_pages: bool,
+    detect_language: bool,
+    load_in_8_bit: bool,
     bin_thresh: float,
     box_thresh: float,
 ):
@@ -130,6 +138,8 @@ def analyze_page(
         reco_arch: recognition architecture
         assume_straight_pages: whether to assume straight pages or not
         straighten_pages: whether to straighten rotated pages or not
+        detect_language: whether to detect the language of the text
+        load_in_8_bit: whether to load the image in 8 bit mode
         bin_thresh: binarization threshold for the segmentation map
         box_thresh: minimal objectness score to consider a box
 
@@ -153,6 +163,8 @@ def analyze_page(
         reco_arch,
         assume_straight_pages,
         straighten_pages,
+        detect_language,
+        load_in_8_bit,
         bin_thresh,
         box_thresh,
     )
@@ -173,8 +185,10 @@ def analyze_page(
 
 
 with gr.Blocks(fill_height=True) as demo:
-    gr.Markdown("# **OnnxTR OCR demo**")
-    gr.Markdown("### This demo showcases the OCR capabilities of OnnxTR. **Github**: [OnnxTR](https://github.com/felixdittrich92/OnnxTR)")
+    #gr.Image("/file=data/logo.jpg", height=400, label=None, show_label=False, show_download_button=False)
+    gr.HTML("<h1>OnnxTR OCR Demo</h1>")
+    gr.HTML("<h2>GitHub: <a href='https://github.com/felixdittrich92/OnnxTR'>OnnxTR</a></h2>")
+    gr.HTML("</div>")
     with gr.Row():
         with gr.Column(scale=1):
             upload = gr.File(label="Upload File [JPG | PNG | PDF]", file_types=["pdf", "jpg", "png"])
@@ -183,6 +197,8 @@ with gr.Blocks(fill_height=True) as demo:
             reco_model = gr.Dropdown(choices=RECO_ARCHS, value=RECO_ARCHS[0], label="Text recognition model")
             assume_straight = gr.Checkbox(value=True, label="Assume straight pages")
             straighten = gr.Checkbox(value=False, label="Straighten pages")
+            det_language = gr.Checkbox(value=False, label="Detect language")
+            load_in_8_bit = gr.Checkbox(value=False, label="Load 8-bit quantized models")
             binarization_threshold = gr.Slider(
                 minimum=0.1, maximum=0.9, value=0.3, step=0.1, label="Binarization threshold"
             )
@@ -208,10 +224,12 @@ with gr.Blocks(fill_height=True) as demo:
             reco_model,
             assume_straight,
             straighten,
+            det_language,
+            load_in_8_bit,
             binarization_threshold,
             box_threshold,
         ],
         outputs=[input_image, segmentation_heatmap, output_image, ocr_output],
     )
 
-demo.launch(inbrowser=True)
+demo.launch(inbrowser=True, allowed_paths=['./data/logo.jpg'])