added output image

app.py

@@ -2,13 +2,13 @@ import gradio as gr
 from detector import bat_detector
 
 def find_bats(input_img):
-    bats, time = bat_detector(input_img)
-    return f"{bats}" + " bats found in " + f"{time:.2f}s"
+    after_image, total_bats = bat_detector(input_img)
+    return f"{total_bats}" + " bats found in " + f"{time:.2f}s", after_image
 
 iface = gr.Interface(
     fn=find_bats, 
     inputs=gr.Image(shape=(640, 512)), 
-    outputs='text'
+    outputs=['text', gr.Image(shape=(960, 768))]
 )
 
 iface.launch()

detector.py

@@ -63,20 +63,39 @@ def find_bats(allImgs):
                                
     return cropped_bats
 
-def bat_detector(img):
-    start=time.time()
-    allImgs = chop_img(img, 3)
-    cropped_bats = find_bats(allImgs)
-
+def find_bats(allImgs, learn):
+    total_bats, batDepthMin,batDepthMax = 0, 50, 400
+    allImgAfter = []
     learn = load_learner("model_densenet_zeros.pkl")
 
-    def predict(bat):
-        with learn.no_bar(), learn.no_logging():
-            start = time.time()
-            _, _, probs = learn.predict(bat)
-            print(f"{time.time()-start:.4f}s")
-        return (tuple(map(lambda x: f"{x:.4f}", probs)))
+    for img in allImgs:
+        blobs = cv.findContours(img[1], cv.RETR_LIST, cv.CHAIN_APPROX_SIMPLE)[-2]
+        after = img[0].copy()
+        # Process the blobs
+        for blob in blobs:
+            if batDepthMin < cv.contourArea(blob) < batDepthMax:  # Only process blobs with a min / max size 
+                rect = cv.minAreaRect(blob)
+                box = cv.boxPoints(rect)
+                box = np.int0(box)
+                cropped_bat = crop_bat(img[0], box)
+                with learn.no_bar(), learn.no_logging():
+                    label, _, probs = learn.predict(cropped_bat)
+                p_not_bat=f"{probs[0]:.4f}"
+                p_bat=f"{probs[1]:.4f}"
+                if p_not_bat < '0.9' and p_bat > '0.5':
+                    cv.drawContours(after, [box], 0, (0,0,255),1)
+                    total_bats += 1
+        allImgAfter.append(after)
+                
+    return allImgAfter, total_bats
 
-    filtered_results = filter(lambda x: (x[0] < '0.9' and x[1] > '0.5'), [predict(bat) for bat in cropped_bats]) # x[0] is !bat probability and x[1] is bat probability
+def bat_detector(img):
+    allImgs = chop_img(img, 3)
+    allImgsAfter, total_bats = find_bats(allImgs)
+    img_row_after_1 = cv.hconcat([allImgsAfter[0],allImgsAfter[1],allImgsAfter[2]])
+    img_row_after_2 = cv.hconcat([allImgsAfter[3],allImgsAfter[4],allImgsAfter[5]])
+    img_row_after_3 = cv.hconcat([allImgsAfter[6],allImgsAfter[7],allImgsAfter[8]])
+    img_concat_after = cv.resize(cv.vconcat([img_row_after_1, img_row_after_2, img_row_after_3]), (960, 768))
 
-    return len(list(filtered_results)), (time.time()-start)
+    return img_concat_after, total_bats
+