add: resize image

hloc/extract_features.py

@@ -73,6 +73,10 @@ confs = {
         "preprocessing": {
             "grayscale": True,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "r2d2": {
@@ -102,6 +106,10 @@ confs = {
         "preprocessing": {
             "grayscale": False,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "d2net-ms": {
@@ -114,6 +122,10 @@ confs = {
         "preprocessing": {
             "grayscale": False,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "rord": {
@@ -126,6 +138,10 @@ confs = {
         "preprocessing": {
             "grayscale": False,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "rootsift": {
@@ -201,6 +217,10 @@ confs = {
         "preprocessing": {
             "grayscale": False,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "xfeat": {
@@ -212,6 +232,10 @@ confs = {
         "preprocessing": {
             "grayscale": False,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "alike": {
@@ -228,6 +252,10 @@ confs = {
         "preprocessing": {
             "grayscale": False,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "lanet": {
@@ -240,6 +268,10 @@ confs = {
         "preprocessing": {
             "grayscale": False,
             "resize_max": 1600,
+            "force_resize": True,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "darkfeat": {

hloc/match_dense.py

@@ -91,7 +91,14 @@ confs = {
             "max_keypoints": 2000,
             "match_threshold": 0.2,
         },
-        "preprocessing": {"grayscale": True, "resize_max": 1024, "dfactor": 8},
+        "preprocessing": {
+            "grayscale": True,
+            "resize_max": 1024,
+            "dfactor": 8,
+            "width": 640,
+            "height": 480,
+            "force_resize": True,
+        },
         "max_error": 2,  # max error for assigned keypoints (in px)
         "cell_size": 8,  # size of quantization patch (max 1 kp/patch)
     },
@@ -104,7 +111,14 @@ confs = {
             "max_keypoints": 2000,
             "match_threshold": 0.2,
         },
-        "preprocessing": {"grayscale": True, "resize_max": 1024, "dfactor": 8},
+        "preprocessing": {
+            "grayscale": True,
+            "resize_max": 1024,
+            "dfactor": 8,
+            "width": 640,
+            "height": 480,
+            "force_resize": True,
+        },
         "max_error": 4,  # max error for assigned keypoints (in px)
         "cell_size": 4,  # size of quantization patch (max 1 kp/patch)
     },
@@ -251,6 +265,10 @@ confs = {
             "resize_max": 1024,
             "dfactor": 8,
             "force_resize": False,
+            "resize_max": 1024,
+            "width": 640,
+            "height": 480,
+            "dfactor": 8,
         },
     },
     "sold2": {

ui/app_class.py

@@ -109,6 +109,29 @@ class ImageMatchingApp:
                             )
 
                         with gr.Accordion("Advanced Setting", open=False):
+                            with gr.Accordion("Image Setting", open=True):
+                                with gr.Row():
+                                    image_force_resize_cb = gr.Checkbox(
+                                        label="Force Resize",
+                                        value=False,
+                                        interactive=True,
+                                    )
+                                    image_setting_height = gr.Slider(
+                                        minimum=48,
+                                        maximum=2048,
+                                        step=16,
+                                        label="Image Height",
+                                        value=480,
+                                        visible=False,
+                                    )
+                                    image_setting_width = gr.Slider(
+                                        minimum=64,
+                                        maximum=2048,
+                                        step=16,
+                                        label="Image Width",
+                                        value=640,
+                                        visible=False,
+                                    )
                             with gr.Accordion("Matching Setting", open=True):
                                 with gr.Row():
                                     match_setting_threshold = gr.Slider(
@@ -195,7 +218,12 @@ class ImageMatchingApp:
                                             "setting_geometry"
                                         ],
                                     )
-
+                        # image resize
+                        image_force_resize_cb.select(
+                            fn=self._on_select_force_resize,
+                            inputs=image_force_resize_cb,
+                            outputs=[image_setting_width, image_setting_height],
+                        )
                         # collect inputs
                         state_cache = gr.State({})
                         inputs = [
@@ -211,7 +239,9 @@ class ImageMatchingApp:
                             ransac_max_iter,
                             choice_geometry_type,
                             gr.State(self.matcher_zoo),
-                            # state_cache,
+                            image_force_resize_cb,
+                            image_setting_width,
+                            image_setting_height,
                         ]
 
                         # Add some examples
@@ -295,7 +325,6 @@ class ImageMatchingApp:
                         inputs=match_image_src,
                         outputs=input_image1,
                     )
-
                     # collect outputs
                     outputs = [
                         output_keypoints,
@@ -336,6 +365,7 @@ class ImageMatchingApp:
                         ransac_max_iter,
                         choice_geometry_type,
                         output_pred,
+                        image_force_resize_cb,
                     ]
                     button_reset.click(
                         fn=self.ui_reset_state,
@@ -422,6 +452,27 @@ class ImageMatchingApp:
                 "source": choice,  # The list of image sources to be displayed
             }
 
+    def _on_select_force_resize(self, visible: bool = False):
+        image_height = gr.Slider(
+            minimum=48,
+            maximum=2048,
+            step=16,
+            label="Image Height",
+            value=480,
+            visible=visible,
+            interactive=True,
+        )
+        image_width = gr.Slider(
+            minimum=64,
+            maximum=2048,
+            step=16,
+            label="Image Width",
+            value=640,
+            visible=visible,
+            interactive=True,
+        )
+        return image_width, image_height
+
     def ui_reset_state(
         self,
         *args: Any,
@@ -446,6 +497,7 @@ class ImageMatchingApp:
         int,
         float,
         int,
+        bool,
     ]:
         """
         Reset the state of the UI.
@@ -487,6 +539,7 @@ class ImageMatchingApp:
             self.cfg["defaults"]["ransac_max_iter"],  # ransac_max_iter: int
             self.cfg["defaults"]["setting_geometry"],  # geometry: str
             None,  # predictions
+            False,
         )
 
     def display_supported_algorithms(self, style="tab"):

ui/utils.py

@@ -829,6 +829,9 @@ def run_matching(
     ransac_max_iter: int = DEFAULT_RANSAC_MAX_ITER,
     choice_geometry_type: str = DEFAULT_SETTING_GEOMETRY,
     matcher_zoo: Dict[str, Any] = None,
+    force_resize: bool = False,
+    image_width: int = 640,
+    image_height: int = 480,
     use_cached_model: bool = False,
 ) -> Tuple[
     np.ndarray,
@@ -902,6 +905,12 @@ def run_matching(
     t1 = time.time()
 
     if model["dense"]:
+        match_conf["preprocessing"]["force_resize"] = force_resize
+        if force_resize:
+            match_conf["preprocessing"]["height"] = image_height
+            match_conf["preprocessing"]["width"] = image_width
+            logger.info(f"Force resize to {image_width}x{image_height}")
+
         pred = match_dense.match_images(
             matcher, image0, image1, match_conf["preprocessing"], device=DEVICE
         )
@@ -925,6 +934,12 @@ def run_matching(
         else:
             extractor = get_feature_model(extract_conf)
 
+        extract_conf["preprocessing"]["force_resize"] = force_resize
+        if force_resize:
+            extract_conf["preprocessing"]["height"] = image_height
+            extract_conf["preprocessing"]["width"] = image_width
+            logger.info(f"Force resize to {image_width}x{image_height}")
+
         pred0 = extract_features.extract(
             extractor, image0, extract_conf["preprocessing"]
         )