[Feature] Enable import/export of ODVG annotations to and from the Gr…

…ounding DINO dataset

anylabeling/resources/resources.qrc

@@ -29,6 +29,7 @@
         <file>images/fit.png</file>
         <file>images/format_classify.png</file>
         <file>images/format_default.png</file>
+        <file>images/format_odvg.png</file>
         <file>images/format_mot.png</file>
         <file>images/format_ppocr.png</file>
         <file>images/format_voc.png</file>

anylabeling/views/labeling/label_converter.py

@@ -2,6 +2,7 @@
 import os.path as osp
 import cv2
 import json
+import jsonlines
 import math
 import yaml
 import pathlib
@@ -720,6 +721,43 @@ def mot_to_custom(self, input_file, output_path, image_path):
             with open(output_file, "w", encoding="utf-8") as f:
                 json.dump(self.custom_data, f, indent=2, ensure_ascii=False)
 
+    def odvg_to_custom(self, input_file, output_path):
+        # Load od.json or od.jsonl
+        with jsonlines.open(input_file, 'r') as reader:
+            od_data = list(reader)
+        # Save custom info
+        for data in od_data:
+            self.reset()
+            shapes = []
+            for instance in data["detection"]["instance"]:
+                xmin, ymin, xmax, ymax = instance["bbox"]
+                points = [
+                    [xmin, ymin],
+                    [xmax, ymin],
+                    [xmax, ymax],
+                    [xmin, ymax],
+                ]
+                shape = {
+                    "label": instance["category"],
+                    "description": None,
+                    "points": points,
+                    "group_id": None,
+                    "difficult": False,
+                    "direction": 0,
+                    "shape_type": "rectangle",
+                    "flags": {},
+                }
+                shapes.append(shape)
+            self.custom_data["imagePath"] = data["filename"]
+            self.custom_data["imageHeight"] = data["height"]
+            self.custom_data["imageWidth"] = data["width"]
+            self.custom_data["shapes"] = shapes
+            output_file = osp.join(
+                output_path, osp.splitext(data["filename"])[0] + ".json"
+            )
+            with open(output_file, "w", encoding="utf-8") as f:
+                json.dump(self.custom_data, f, indent=2, ensure_ascii=False)
+
     def ppocr_to_custom(self, input_file, output_path, image_path, mode):
         if mode in ["rec", "kie"]:
             with open(input_file, "r", encoding="utf-8") as f:
@@ -1252,6 +1290,54 @@ def custom_to_mot(self, input_path, save_path):
             for row in mot_structure["gt"]:
                 f.write(",".join(map(str, row)) + "\n")
 
+    def custom_to_odvg(self, image_list, label_path, save_path):
+        # Save label_map.json
+        label_map = {}
+        for i, c in enumerate(self.classes):
+            label_map[i] = c
+        label_map_file = osp.join(save_path, "label_map.json")
+        with open(label_map_file, 'w') as f:
+            json.dump(label_map, f)
+        # Save od.json
+        od_data = []
+        for image_file in image_list:
+            image_name = osp.basename(image_file)
+            label_name = osp.splitext(image_name)[0] + ".json"
+            label_file = osp.join(label_path, label_name)
+            img = cv2.imdecode(np.fromfile(image_file, dtype=np.uint8), 1)
+            height, width = img.shape[:2]
+            with open(label_file, "r", encoding="utf-8") as f:
+                data = json.load(f)
+            instance = []
+            for shape in data["shapes"]:
+                if (shape["shape_type"] != "rectangle"
+                    or shape["label"] not in self.classes):
+                    continue
+                points = shape["points"]
+                xmin = float(points[0][0])
+                ymin = float(points[0][1])
+                xmax = float(points[2][0])
+                ymax = float(points[2][1])
+                bbox = [xmin, ymin, xmax, ymax]
+                label = self.classes.index(shape["label"])
+                category = shape["label"]
+                instance.append({
+                    "bbox": bbox,
+                    "label": label,
+                    "category": category
+                })
+            od_data.append({
+                "filename": image_name,
+                "height": height,
+                "width": width,
+                "detection": {
+                    "instance": instance
+                }
+            })
+        od_file = osp.join(save_path, "od.json")
+        with jsonlines.open(od_file, mode='w') as writer:
+            writer.write_all(od_data)
+
     def custom_to_pporc(self, image_file, label_file, save_path, mode):
         if not osp.exists(label_file):
             return

anylabeling/views/labeling/label_widget.py

@@ -951,6 +951,13 @@ def __init__(
             icon="format_mot",
             tip=self.tr("Upload Custom Multi-Object-Tracking Annotations"),
         )
+        upload_odvg_annotation = action(
+            self.tr("&Upload ODVG Annotations"),
+            self.upload_odvg_annotation,
+            None,
+            icon="format_odvg",
+            tip=self.tr("Upload Custom Object Detection Visual Grounding Annotations"),
+        )
         upload_ppocr_rec_annotation = action(
             self.tr("&Upload PPOCR-Rec Annotations"),
             lambda: self.upload_ppocr_annotation("rec"),
@@ -1048,6 +1055,13 @@ def __init__(
             icon="format_mot",
             tip=self.tr("Export Custom Multi-Object-Tracking Annotations"),
         )
+        export_odvg_annotation = action(
+            self.tr("&Export ODVG Annotations"),
+            self.export_odvg_annotation,
+            None,
+            icon="format_odvg",
+            tip=self.tr("Export Custom Object Detection Visual Grounding Annotations"),
+        )
         export_pporc_rec_annotation = action(
             self.tr("&Export PPOCR-Rec Annotations"),
             lambda: self.export_pporc_annotation("rec"),
@@ -1164,6 +1178,7 @@ def __init__(
             upload_dota_annotation=upload_dota_annotation,
             upload_mask_annotation=upload_mask_annotation,
             upload_mot_annotation=upload_mot_annotation,
+            upload_odvg_annotation=upload_odvg_annotation,
             upload_ppocr_rec_annotation=upload_ppocr_rec_annotation,
             upload_ppocr_kie_annotation=upload_ppocr_kie_annotation,
             export_yolo_hbb_annotation=export_yolo_hbb_annotation,
@@ -1177,6 +1192,7 @@ def __init__(
             export_dota_annotation=export_dota_annotation,
             export_mask_annotation=export_mask_annotation,
             export_mot_annotation=export_mot_annotation,
+            export_odvg_annotation=export_odvg_annotation,
             export_pporc_rec_annotation=export_pporc_rec_annotation,
             export_pporc_kie_annotation=export_pporc_kie_annotation,
             zoom=zoom,
@@ -1351,6 +1367,7 @@ def __init__(
                 upload_dota_annotation,
                 upload_mask_annotation,
                 upload_mot_annotation,
+                upload_odvg_annotation,
                 None,
                 upload_ppocr_rec_annotation,
                 upload_ppocr_kie_annotation,
@@ -1373,6 +1390,7 @@ def __init__(
                 export_dota_annotation,
                 export_mask_annotation,
                 export_mot_annotation,
+                export_odvg_annotation,
                 None,
                 export_pporc_rec_annotation,
                 export_pporc_kie_annotation,
@@ -4185,12 +4203,6 @@ def upload_mot_annotation(self, _value=False, dirpath=None):
             return
 
         if not self.filename:
-            QtWidgets.QMessageBox.warning(
-                self,
-                self.tr("Warning"),
-                self.tr("Please load an image folder before proceeding!"),
-                QtWidgets.QMessageBox.Ok,
-            )
             return
 
         filter = "Classes Files (*.txt);;All Files (*)"
@@ -4201,12 +4213,6 @@ def upload_mot_annotation(self, _value=False, dirpath=None):
             filter,
         )
         if not self.classes_file:
-            QtWidgets.QMessageBox.warning(
-                self,
-                self.tr("Warning"),
-                self.tr("Please select a specific classes file!"),
-                QtWidgets.QMessageBox.Ok,
-            )
             return
         with open(self.classes_file, "r", encoding="utf-8") as f:
             labels = f.read().splitlines()
@@ -4255,6 +4261,47 @@ def upload_mot_annotation(self, _value=False, dirpath=None):
         # update and refresh the current canvas
         self.load_file(self.filename)
 
+    def upload_odvg_annotation(self, _value=False, dirpath=None):
+        if not self.may_continue():
+            return
+
+        if not self.filename:
+            return
+
+        filter = "OD Files (*.json *.jsonl);;All Files (*)"
+        input_file, _ = QtWidgets.QFileDialog.getOpenFileName(
+            self,
+            self.tr("Select a specific OD file"),
+            "",
+            filter,
+        )
+
+        if (
+            not input_file
+            or QtWidgets.QMessageBox.warning(
+                self,
+                self.tr("Current annotation will be lost"),
+                self.tr(
+                    "You are going to upload new annotations to this task. Continue?"
+                ),
+                QtWidgets.QMessageBox.Cancel | QtWidgets.QMessageBox.Ok,
+            )
+            != QtWidgets.QMessageBox.Ok
+        ):
+            return
+
+        output_dir_path = osp.dirname(self.filename)
+        if self.output_dir:
+            output_dir_path = self.output_dir
+        converter = LabelConverter()
+        converter.odvg_to_custom(
+            input_file=input_file,
+            output_path=output_dir_path,
+        )
+
+        # update and refresh the current canvas
+        self.load_file(self.filename)
+
     def upload_ppocr_annotation(self, mode, _value=False, dirpath=None):
         if not self.may_continue():
             return
@@ -4909,6 +4956,69 @@ def export_mot_annotation(self, _value=False, dirpath=None):
             )
             return
 
+    def export_odvg_annotation(self, _value=False, dirpath=None):
+        if not self.may_continue():
+            return
+
+        if not self.filename:
+            QtWidgets.QMessageBox.warning(
+                self,
+                self.tr("Warning"),
+                self.tr("Please load an image folder before proceeding!"),
+                QtWidgets.QMessageBox.Ok,
+            )
+            return
+
+        if not self.classes_file:
+            filter = "Classes Files (*.txt);;All Files (*)"
+            self.classes_file, _ = QtWidgets.QFileDialog.getOpenFileName(
+                self,
+                self.tr("Select a specific classes file"),
+                "",
+                filter,
+            )
+            if not self.classes_file:
+                return
+
+        label_dir_path = osp.dirname(self.filename)
+        if self.output_dir:
+            label_dir_path = self.output_dir
+
+        selected_dir = QtWidgets.QFileDialog.getExistingDirectory(
+            self,
+            self.tr("Select a directory to save the odvg annotations"),
+            label_dir_path,
+            QtWidgets.QFileDialog.ShowDirsOnly
+        )
+        if not selected_dir:
+            return
+
+        save_path = osp.realpath(selected_dir)
+        image_list = self.image_list
+        if not image_list:
+            image_list = [self.filename]
+        os.makedirs(save_path, exist_ok=True)
+        converter = LabelConverter(classes_file=self.classes_file)
+        try:
+            converter.custom_to_odvg(image_list, label_dir_path, save_path)
+            QtWidgets.QMessageBox.information(
+                self,
+                self.tr("Success"),
+                self.tr(
+                    f"Annotation exported successfully!\n"
+                    f"Check the results in: {save_path}."
+                ),
+                QtWidgets.QMessageBox.Ok,
+            )
+        except Exception as e:
+            QtWidgets.QMessageBox.warning(
+                self,
+                self.tr("Error"),
+                self.tr(f"{e}"),
+                QtWidgets.QMessageBox.Ok,
+            )
+            return
+
     def export_pporc_annotation(self, mode, _value=False, dirpath=None):
         if not self.may_continue():
             return

assets/ODVG/label_map.json

@@ -0,0 +1 @@
+{"0": "person", "1": "bicycle", "2": "car", "3": "motorcycle", "4": "airplane", "5": "bus", "6": "train", "7": "truck", "8": "boat", "9": "traffic light", "10": "fire hydrant", "11": "stop sign", "12": "parking meter", "13": "bench", "14": "bird", "15": "cat", "16": "dog", "17": "horse", "18": "sheep", "19": "cow", "20": "elephant", "21": "bear", "22": "zebra", "23": "giraffe", "24": "backpack", "25": "umbrella", "26": "handbag", "27": "tie", "28": "suitcase", "29": "frisbee", "30": "skis", "31": "snowboard", "32": "sports ball", "33": "kite", "34": "baseball bat", "35": "baseball glove", "36": "skateboard", "37": "surfboard", "38": "tennis racket", "39": "bottle", "40": "wine glass", "41": "cup", "42": "fork", "43": "knife", "44": "spoon", "45": "bowl", "46": "banana", "47": "apple", "48": "sandwich", "49": "orange", "50": "broccoli", "51": "carrot", "52": "hot dog", "53": "pizza", "54": "donut", "55": "cake", "56": "chair", "57": "sofa", "58": "pottedplant", "59": "bed", "60": "diningtable", "61": "toilet", "62": "tvmonitor", "63": "laptop", "64": "mouse", "65": "remote", "66": "keyboard", "67": "cell phone", "68": "microwave", "69": "oven", "70": "toaster", "71": "sink", "72": "refrigerator", "73": "book", "74": "clock", "75": "vase", "76": "scissors", "77": "teddy bear", "78": "hair drier", "79": "toothbrush"}

assets/ODVG/od.json

@@ -0,0 +1 @@
+{"filename": "demo.jpg", "height": 1944, "width": 2592, "detection": {"instance": [{"bbox": [568.3057319641114, 1435.8236297607423, 801.2323951721191, 1940.6873748779299], "label": 0, "category": "person"}, {"bbox": [250.73743057250977, 88.3039478302002, 642.3008560180665, 416.0265106201172], "label": 5, "category": "bus"}, {"bbox": [2044.598277282715, 961.5066833496094, 2592.0, 1443.4265258789064], "label": 2, "category": "car"}, {"bbox": [2168.0969512939455, 291.9082489013672, 2590.0523712158206, 632.3925201416016], "label": 7, "category": "truck"}, {"bbox": [935.0150756835938, 467.1713562011719, 1389.4009002685548, 709.4285430908203], "label": 2, "category": "car"}, {"bbox": [1490.3331344604494, 210.9319793701172, 1983.798097229004, 467.6459655761719], "label": 7, "category": "truck"}, {"bbox": [2474.505505371094, 517.0932792663574, 2591.8625610351564, 759.1917068481446], "label": 2, "category": "car"}, {"bbox": [1387.124382019043, 491.3653106689453, 1790.5220809936525, 701.9515914916992], "label": 2, "category": "car"}, {"bbox": [681.8321090698242, 577.1008369445801, 1127.0446243286133, 879.6766113281251], "label": 2, "category": "car"}, {"bbox": [1701.209248352051, 1034.6568420410156, 1909.2545150756837, 1272.3461608886719], "label": 1, "category": "bicycle"}, {"bbox": [1577.2593246459962, 1287.5478744506836, 1782.4759689331056, 1750.773422241211], "label": 0, "category": "person"}, {"bbox": [341.1669548034668, 837.0524871826173, 495.9798980712891, 1196.966326904297], "label": 0, "category": "person"}, {"bbox": [19.24847688674927, 1068.9359985351564, 176.856941986084, 1450.8116455078125], "label": 0, "category": "person"}, {"bbox": [867.1709632873535, 1638.5384399414063, 1039.1333724975586, 1939.5025817871094], "label": 0, "category": "person"}, {"bbox": [1730.0543746948242, 859.913703918457, 1904.4883987426758, 1192.5239089965821], "label": 0, "category": "person"}, {"bbox": [700.1718063354492, 302.9357482910156, 921.9890258789063, 439.0647171020508], "label": 2, "category": "car"}, {"bbox": [1464.6719696044922, 670.2848258972168, 1594.598016357422, 838.7980361938477], "label": 1, "category": "bicycle"}, {"bbox": [1053.4611373901369, 1550.143803405762, 1263.1692672729494, 1937.8229095458985], "label": 0, "category": "person"}, {"bbox": [76.19504356384277, 262.1977020263672, 340.8912117004395, 425.138639831543], "label": 2, "category": "car"}, {"bbox": [496.75765800476074, 811.980258178711, 632.1751762390137, 1106.7453094482423], "label": 0, "category": "person"}, {"bbox": [483.645276260376, 424.18151092529297, 856.9735977172852, 732.0099380493165], "label": 2, "category": "car"}, {"bbox": [1771.154187011719, 561.9675956726074, 2408.6232971191407, 918.0072509765625], "label": 2, "category": "car"}, {"bbox": [915.2098983764649, 1255.077053833008, 1116.117855834961, 1691.7488250732422], "label": 0, "category": "person"}, {"bbox": [136.0616397857666, 792.3811157226563, 250.77661056518556, 1102.5672637939454], "label": 0, "category": "person"}, {"bbox": [1473.6370193481446, 565.5210479736328, 1589.4288528442385, 795.7627075195313], "label": 0, "category": "person"}, {"bbox": [832.0695831298829, 970.7194198608399, 998.2965728759766, 1371.3438674926758], "label": 0, "category": "person"}, {"bbox": [571.2070289611817, 1130.0795288085938, 720.904387664795, 1497.7851196289064], "label": 0, "category": "person"}, {"bbox": [1071.136134338379, 182.94281845092775, 1350.3359756469727, 361.1679084777832], "label": 2, "category": "car"}, {"bbox": [1541.1687423706055, 396.60825119018557, 1971.5012649536134, 587.4710517883301], "label": 2, "category": "car"}, {"bbox": [960.4221267700195, 348.6872268676758, 1279.5368637084962, 502.6705307006836], "label": 2, "category": "car"}, {"bbox": [325.6000144958496, 349.9100875854492, 389.2794296264649, 518.6988555908204], "label": 0, "category": "person"}, {"bbox": [1543.162843322754, 397.4399917602539, 1965.3402420043947, 590.7352890014648], "label": 7, "category": "truck"}, {"bbox": [1222.1846466064453, 970.3025299072266, 1342.1696044921875, 1370.0215118408205], "label": 0, "category": "person"}, {"bbox": [1808.9995880126953, 352.1537910461426, 2187.0086517333984, 556.4679359436035], "label": 7, "category": "truck"}, {"bbox": [265.45752067565917, 688.9645980834962, 358.0454669952393, 991.6645248413087], "label": 0, "category": "person"}, {"bbox": [626.5533416748048, 755.5319000244141, 750.7314239501953, 1082.483624267578], "label": 0, "category": "person"}, {"bbox": [1193.545555114746, 1233.8073852539064, 1256.292869567871, 1315.7788513183596], "label": 26, "category": "handbag"}, {"bbox": [1541.7802963256836, 1533.2621704101564, 1653.9361770629885, 1689.3923400878907], "label": 26, "category": "handbag"}, {"bbox": [1180.547314453125, 1387.6520141601563, 1259.1935485839845, 1533.373406982422], "label": 26, "category": "handbag"}, {"bbox": [14.264483642578126, 600.0972061157227, 110.88783016204835, 860.7037307739258], "label": 0, "category": "person"}, {"bbox": [345.31243286132815, 994.7911376953125, 394.4514976501465, 1116.7568481445312], "label": 26, "category": "handbag"}, {"bbox": [461.77211837768556, 608.4305557250976, 561.3444854736329, 825.7744583129884], "label": 0, "category": "person"}, {"bbox": [0.0, 894.918864440918, 77.95618114471436, 1284.5674758911134], "label": 0, "category": "person"}, {"bbox": [141.36901473999023, 346.4582313537598, 218.52517318725586, 523.6349784851075], "label": 0, "category": "person"}, {"bbox": [76.45697479248047, 1147.9113693237305, 140.87081394195556, 1236.9033462524415], "label": 26, "category": "handbag"}, {"bbox": [0.7425890922546388, 157.6424617767334, 75.8009262084961, 316.4888122558594], "label": 2, "category": "car"}, {"bbox": [1821.8278839111329, 350.3739440917969, 2190.6821777343753, 559.7829711914063], "label": 2, "category": "car"}, {"bbox": [73.8185359954834, 994.9901275634766, 140.259352684021, 1073.2087188720704], "label": 26, "category": "handbag"}, {"bbox": [1294.401530456543, 1097.295021057129, 1370.207276916504, 1230.059083557129], "label": 26, "category": "handbag"}]}}

requirements-gpu.txt

@@ -14,4 +14,5 @@ scipy
 shapely
 pyclipper
 filterpy
-tokenizers
+tokenizers
+jsonlines

requirements-macos.txt

@@ -15,4 +15,5 @@ scipy
 shapely
 pyclipper
 filterpy
-tokenizers
+tokenizers
+jsonlines

requirements.txt

@@ -14,4 +14,5 @@ scipy
 shapely
 pyclipper
 filterpy
-tokenizers
+tokenizers
+jsonlines