Update app.py

app.py

@@ -179,12 +179,12 @@ ocr_id = {
 }
 
 
-def blur_im(img,bounds,target_lang,trans_lang):
+def blur_im(img,bounds,target_lang,trans_lang,ocr_sens,font_fac):
     im = cv2.imread(img)
     im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB)
    
     for bound in bounds:
-        if bound[2]>=0.3:
+        if bound[2]>=(ocr_sens):
             p0, p1, p2, p3 = bound[0]
             x = int(p0[0])
             y = int(p0[1])
@@ -197,7 +197,7 @@ def blur_im(img,bounds,target_lang,trans_lang):
             pass
     im = Image.fromarray(im)
     for bound in bounds:
-        if bound[2]>=0.3:
+        if bound[2]>=(ocr_sens):
             p0, p1, p2, p3 = bound[0]
             x = int(p0[0])
             y = int(p0[1])
@@ -205,20 +205,17 @@ def blur_im(img,bounds,target_lang,trans_lang):
             h = int(p2[1]) - int(y)
             draw = ImageDraw.Draw(im)
             text = this(bound[1],target_lang,trans_lang)
-          
-            font_size=int(int(w)*0.1)
-
+            font_size=int(int(h)*font_fac)
             font = ImageFont.truetype("./fonts/unifont-15.0.01.ttf", int(font_size))
-                    
             draw.text((x, y),text, font = font, fill=(0,0,0))
         else:
             pass
     return im
     
-def draw_boxes(image, bounds, width=1):
+def draw_boxes(image, bounds, ocr_sens,width=1):
     draw = ImageDraw.Draw(image)
     for bound in bounds:
-        if bound[2]>=0.3:
+        if bound[2]>=(ocr_sens):
             color = "blue"
         else:
             color = "red"
@@ -226,7 +223,7 @@ def draw_boxes(image, bounds, width=1):
         draw.line([*p0, *p1, *p2, *p3, *p0], fill=color, width=width)
     return image
 
-def detect(img, target_lang,trans_lang,target_lang2=None):
+def detect(img, target_lang,trans_lang,ocr_sens,font_fac,target_lang2=None):
     if target_lang2 != None and target_lang2 != "":
         lang=f"{lang_id[target_lang]}"
         lang2=f"{lang_id[target_lang2]}"
@@ -234,28 +231,33 @@ def detect(img, target_lang,trans_lang,target_lang2=None):
     else:
         lang=[f"{ocr_id[target_lang]}"]
         pass
-    #img = Image.open(img)
 
-    #img.thumbnail((1000,1000))
-    #img = np.array(img)
+    img = Image.open(img)
+    img.thumbnail((1000,1000), Image.Resampling.LANCZOS)
+    img.save("tmp.jpg")
+    img1 = np.array(img)
     reader = easyocr.Reader(lang)
-    bounds = reader.readtext(img)
-    
-    im = PIL.Image.open(img)
-    im_out=draw_boxes(im, bounds)
-    #im.save('result.jpg')
-
-    blr_out=blur_im(img,bounds,target_lang,trans_lang)
+    bounds = reader.readtext(img1)
+    im = PIL.Image.open("tmp.jpg")
+    im_out=draw_boxes(im, bounds,ocr_sens)
+    blr_out=blur_im("tmp.jpg",bounds,target_lang,trans_lang,ocr_sens,font_fac)
     return im_out,blr_out,pd.DataFrame(bounds),pd.DataFrame(bounds).iloc[:,1:]
     
 with gr.Blocks() as robot:
     with gr.Row():
         with gr.Column():
             im=gr.Image(type="filepath")
+
         with gr.Column():
-            target_lang = gr.Dropdown(label="Detect language:", choices=list(ocr_id.keys()),value="English")
-            #target_lang2 = gr.Dropdown(label="Detect language2", choices=list(lang_id.keys()),value="")
-            trans_lang = gr.Dropdown(label="Translate to:", choices=list(lang_id.keys()),value="Chinese")
+            with gr.Group():
+                with gr.Row():
+                    with gr.Column():
+                        target_lang = gr.Dropdown(label="Detect language:", choices=list(ocr_id.keys()),value="English")
+                        trans_lang = gr.Dropdown(label="Translate to:", choices=list(lang_id.keys()),value="Chinese")
+                    with gr.Column():
+                        ocr_sens=gr.Slider(0.1, 1, step=0.05,value=0.25,label="Detect Min Confidence")
+                        font_fac=gr.Slider(0.1, 1, step =0.1,value=0.4,label="Font Scale")
+                        ocr_space=gr.Slider(1,10, step=1,value=5,label="Future Function")            
             go_btn=gr.Button()
     with gr.Row():
         with gr.Column():
@@ -267,6 +269,6 @@ with gr.Blocks() as robot:
         out_txt=gr.Textbox(lines=8)
         data_f=gr.Dataframe()
                         
-    go_btn.click(detect,[im,target_lang,trans_lang],[out_im,trans_im,out_txt,data_f])
+    go_btn.click(detect,[im,target_lang,trans_lang,ocr_sens,font_fac],[out_im,trans_im,out_txt,data_f])
     #go_btn.click(detect,[im,target_lang,target_lang2],[out_im,trans_im,out_txt,data_f])
 robot.queue(concurrency_count=10).launch()