Upload app.py

app.py

@@ -1052,10 +1052,12 @@ def prepare_export(stored_masks, stored_image, threshold_bias):
 # Tab builder
 # ---------------------------------------------------------------------------
 
-def draw_polygon_overlay(image_pil, points):
+def draw_polygon_overlay(image_pil, points, stem_px=0):
     """
     Draw numbered vertex dots and polygon edges onto a copy of image_pil.
-    points: list of (x, y) tuples in original image pixel space.
+    points: list of (x, y) tuples in preview pixel space.
+    stem_px: if non-zero, draw a tail from the last vertex down to the touch
+             point it was offset from (see add_crop_point).
     Returns a new PIL image.
     """
     img = image_pil.copy().convert("RGBA")
@@ -1078,6 +1080,15 @@ def draw_polygon_overlay(image_pil, points):
                      fill=(74, 170, 255, 255), outline=(255, 255, 255, 255))
         draw.text((x, y), str(i + 1), fill=(255, 255, 255, 255), anchor="mm")
 
+    # In offset mode the marker sits above the fingertip. Without something
+    # connecting the two it reads as "the app put my point in the wrong place",
+    # so draw the stem down to where the touch actually landed.
+    if stem_px and points:
+        x, y = points[-1]
+        draw.line([(x, y + r), (x, y + stem_px)], fill=(74, 170, 255, 140), width=2)
+        draw.ellipse([x - 5, y + stem_px - 5, x + 5, y + stem_px + 5],
+                     outline=(255, 255, 255, 170), width=2)
+
     combined = Image.alpha_composite(img, overlay)
     return combined.convert("RGB")
 
@@ -1100,7 +1111,10 @@ def make_preview(image_pil):
     return preview, (preview.width / float(image_pil.width)) if image_pil.width else 1.0
 
 
-def add_crop_point(preview_pil, points, scale, evt: gr.SelectData):
+OFFSET_PCT_DEFAULT = 10.0   # % of picker height; ~a fingertip on a phone
+
+
+def add_crop_point(preview_pil, points, scale, offset_pct, evt: gr.SelectData):
     """
     Called by gr.Image .select(). Appends the clicked coordinate,
     redraws the overlay on the PREVIEW, returns (updated_preview, updated_points).
@@ -1118,17 +1132,27 @@ def add_crop_point(preview_pil, points, scale, evt: gr.SelectData):
 
     if preview_pil is None:
         return preview_pil, points
+
+    # Vertical offset so the marker is not hidden under the fingertip.
+    # Expressed as a share of picker height, which keeps it consistent across
+    # phone sizes instead of being tied to the image's pixel dimensions.
+    stem = int(preview_pil.height * (float(offset_pct or 0.0) / 100.0))
+
     if len(points) >= 4:
-        return draw_polygon_overlay(preview_pil, to_preview(points)), points
+        return draw_polygon_overlay(preview_pil, to_preview(points), stem), points
 
     # Gradio can deliver an empty selection (see gradio-app/gradio#5945);
     # dropping it beats raising a red error box the user cannot interpret.
     if evt is None or evt.index is None or evt.index[0] is None:
         return draw_polygon_overlay(preview_pil, to_preview(points)), points
 
-    x, y = int(evt.index[0] / scale), int(evt.index[1] / scale)
+    tx, ty = int(evt.index[0]), int(evt.index[1])
+    # Lift the marker above the touch, clamped so a tap near the top edge
+    # cannot place a vertex off the image.
+    my = max(0, min(preview_pil.height - 1, ty - stem))
+    x, y = int(tx / scale), int(my / scale)
     new_points = points + [(x, y)]
-    return draw_polygon_overlay(preview_pil, to_preview(new_points)), new_points
+    return draw_polygon_overlay(preview_pil, to_preview(new_points), stem), new_points
 
 
 def clear_crop_points(image_pil):
@@ -1316,8 +1340,10 @@ def build_tab(tab_index, masks_state, image_state, result_state):
 
                 gr.Markdown(
                     "### Crop region (optional)\n"
-                    "Click/tap up to **4 points** on the image below to define the region "
-                    "to segment. The polygon will be drawn as you click. "
+                    "Tap up to **4 points** on the image below to define the region "
+                    "to segment. On a touch screen the marker is placed **above "
+                    "your fingertip** so it is not hidden — a faint tail shows "
+                    "where you actually touched. "
                     "Leave empty to segment the full image."
                 )
 
@@ -1334,6 +1360,13 @@ def build_tab(tab_index, masks_state, image_state, result_state):
 
                 clear_crop_btn = gr.Button("✕ Clear crop points", size="sm")
 
+                touch_offset = gr.Slider(
+                    minimum=0, maximum=25, value=OFFSET_PCT_DEFAULT, step=1,
+                    label="Fingertip offset (% of picker height)",
+                    info="How far above your finger the marker is placed. "
+                         "Set to 0 when using a mouse.",
+                )
+
                 model_dropdown = gr.Dropdown(
                     choices=list(MODEL_OPTIONS.keys()),
                     label="Select Model",
@@ -1520,9 +1553,10 @@ def build_tab(tab_index, masks_state, image_state, result_state):
             _component.change(fn=on_exclusion_change,
                 inputs=exclusion_inputs, outputs=exclusion_outputs)
 
-        def on_crop_click(base_img, points, scale, evt: gr.SelectData):
+        def on_crop_click(base_img, points, scale, offset_pct, evt: gr.SelectData):
             before = len(points or [])
-            updated_img, updated_pts = add_crop_point(base_img, points, scale, evt)
+            updated_img, updated_pts = add_crop_point(base_img, points, scale,
+                                                      offset_pct, evt)
             n = len(updated_pts)
             if n == before and n < 4:
                 status = "*Tap not registered — try again slightly inside the image*"
@@ -1533,7 +1567,8 @@ def build_tab(tab_index, masks_state, image_state, result_state):
             return updated_img, updated_pts, status
 
         crop_display.select(fn=on_crop_click,
-            inputs=[base_image_state, crop_points_state, preview_scale_state],
+            inputs=[base_image_state, crop_points_state, preview_scale_state,
+                    touch_offset],
             outputs=[crop_display, crop_points_state, crop_status])
 
         def on_clear_crop(base_img):