add files

GroundingDINO/groundingdino/config/GroundingDINO_SwinB.cfg.py

@@ -0,0 +1,43 @@
+batch_size = 1
+modelname = "groundingdino"
+backbone = "swin_B_384_22k"
+position_embedding = "sine"
+pe_temperatureH = 20
+pe_temperatureW = 20
+return_interm_indices = [1, 2, 3]
+backbone_freeze_keywords = None
+enc_layers = 6
+dec_layers = 6
+pre_norm = False
+dim_feedforward = 2048
+hidden_dim = 256
+dropout = 0.0
+nheads = 8
+num_queries = 900
+query_dim = 4
+num_patterns = 0
+num_feature_levels = 4
+enc_n_points = 4
+dec_n_points = 4
+two_stage_type = "standard"
+two_stage_bbox_embed_share = False
+two_stage_class_embed_share = False
+transformer_activation = "relu"
+dec_pred_bbox_embed_share = True
+dn_box_noise_scale = 1.0
+dn_label_noise_ratio = 0.5
+dn_label_coef = 1.0
+dn_bbox_coef = 1.0
+embed_init_tgt = True
+dn_labelbook_size = 2000
+max_text_len = 256
+text_encoder_type = "bert-base-uncased"
+use_text_enhancer = True
+use_fusion_layer = True
+use_checkpoint = True
+use_transformer_ckpt = True
+use_text_cross_attention = True
+text_dropout = 0.0
+fusion_dropout = 0.0
+fusion_droppath = 0.1
+sub_sentence_present = True

GroundingDINO/groundingdino/util/inference.py

@@ -13,6 +13,10 @@ from groundingdino.util.misc import clean_state_dict
 from groundingdino.util.slconfig import SLConfig
 from groundingdino.util.utils import get_phrases_from_posmap
 
+# ----------------------------------------------------------------------------------------------------------------------
+# OLD API
+# ----------------------------------------------------------------------------------------------------------------------
+
 
 def preprocess_caption(caption: str) -> str:
     result = caption.lower().strip()
@@ -96,3 +100,143 @@ def annotate(image_source: np.ndarray, boxes: torch.Tensor, logits: torch.Tensor
     annotated_frame = cv2.cvtColor(image_source, cv2.COLOR_RGB2BGR)
     annotated_frame = box_annotator.annotate(scene=annotated_frame, detections=detections, labels=labels)
     return annotated_frame
+
+
+# ----------------------------------------------------------------------------------------------------------------------
+# NEW API
+# ----------------------------------------------------------------------------------------------------------------------
+
+
+class Model:
+
+    def __init__(
+        self,
+        model_config_path: str,
+        model_checkpoint_path: str,
+        device: str = "cuda"
+    ):
+        self.model = load_model(
+            model_config_path=model_config_path,
+            model_checkpoint_path=model_checkpoint_path,
+            device=device
+        ).to(device)
+        self.device = device
+
+    def predict_with_caption(
+        self,
+        image: np.ndarray,
+        caption: str,
+        box_threshold: float = 0.35,
+        text_threshold: float = 0.25
+    ) -> Tuple[sv.Detections, List[str]]:
+        """
+        import cv2
+
+        image = cv2.imread(IMAGE_PATH)
+
+        model = Model(model_config_path=CONFIG_PATH, model_checkpoint_path=WEIGHTS_PATH)
+        detections, labels = model.predict_with_caption(
+            image=image,
+            caption=caption,
+            box_threshold=BOX_THRESHOLD,
+            text_threshold=TEXT_THRESHOLD
+        )
+
+        import supervision as sv
+
+        box_annotator = sv.BoxAnnotator()
+        annotated_image = box_annotator.annotate(scene=image, detections=detections, labels=labels)
+        """
+        processed_image = Model.preprocess_image(image_bgr=image).to(self.device)
+        boxes, logits, phrases = predict(
+            model=self.model,
+            image=processed_image,
+            caption=caption,
+            box_threshold=box_threshold,
+            text_threshold=text_threshold)
+        source_h, source_w, _ = image.shape
+        detections = Model.post_process_result(
+            source_h=source_h,
+            source_w=source_w,
+            boxes=boxes,
+            logits=logits)
+        return detections, phrases
+
+    def predict_with_classes(
+        self,
+        image: np.ndarray,
+        classes: List[str],
+        box_threshold: float,
+        text_threshold: float
+    ) -> sv.Detections:
+        """
+        import cv2
+
+        image = cv2.imread(IMAGE_PATH)
+
+        model = Model(model_config_path=CONFIG_PATH, model_checkpoint_path=WEIGHTS_PATH)
+        detections = model.predict_with_classes(
+            image=image,
+            classes=CLASSES,
+            box_threshold=BOX_THRESHOLD,
+            text_threshold=TEXT_THRESHOLD
+        )
+
+
+        import supervision as sv
+
+        box_annotator = sv.BoxAnnotator()
+        annotated_image = box_annotator.annotate(scene=image, detections=detections)
+        """
+        caption = ", ".join(classes)
+        processed_image = Model.preprocess_image(image_bgr=image).to(self.device)
+        boxes, logits, phrases = predict(
+            model=self.model,
+            image=processed_image,
+            caption=caption,
+            box_threshold=box_threshold,
+            text_threshold=text_threshold)
+        source_h, source_w, _ = image.shape
+        detections = Model.post_process_result(
+            source_h=source_h,
+            source_w=source_w,
+            boxes=boxes,
+            logits=logits)
+        class_id = Model.phrases2classes(phrases=phrases, classes=classes)
+        detections.class_id = class_id
+        return detections
+
+    @staticmethod
+    def preprocess_image(image_bgr: np.ndarray) -> torch.Tensor:
+        transform = T.Compose(
+            [
+                T.RandomResize([800], max_size=1333),
+                T.ToTensor(),
+                T.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
+            ]
+        )
+        image_pillow = Image.fromarray(cv2.cvtColor(image_bgr, cv2.COLOR_BGR2RGB))
+        image_transformed, _ = transform(image_pillow, None)
+        return image_transformed
+
+    @staticmethod
+    def post_process_result(
+            source_h: int,
+            source_w: int,
+            boxes: torch.Tensor,
+            logits: torch.Tensor
+    ) -> sv.Detections:
+        boxes = boxes * torch.Tensor([source_w, source_h, source_w, source_h])
+        xyxy = box_convert(boxes=boxes, in_fmt="cxcywh", out_fmt="xyxy").numpy()
+        confidence = logits.numpy()
+        return sv.Detections(xyxy=xyxy, confidence=confidence)
+
+    @staticmethod
+    def phrases2classes(phrases: List[str], classes: List[str]) -> np.ndarray:
+        class_ids = []
+        for phrase in phrases:
+            try:
+                class_ids.append(classes.index(phrase))
+            except ValueError:
+                class_ids.append(None)
+        return np.array(class_ids)

GroundingDINO/groundingdino/util/slconfig.py

@@ -2,13 +2,13 @@
 # Modified from mmcv
 # ==========================================================
 import ast
+import os
 import os.path as osp
 import shutil
 import sys
 import tempfile
 from argparse import Action
 from importlib import import_module
-import platform
 
 from addict import Dict
 from yapf.yapflib.yapf_api import FormatCode
@@ -81,7 +81,7 @@ class SLConfig(object):
             with tempfile.TemporaryDirectory() as temp_config_dir:
                 temp_config_file = tempfile.NamedTemporaryFile(dir=temp_config_dir, suffix=".py")
                 temp_config_name = osp.basename(temp_config_file.name)
-                if platform.system() == 'Windows':
+                if os.name == 'nt':
                     temp_config_file.close()
                 shutil.copyfile(filename, osp.join(temp_config_dir, temp_config_name))
                 temp_module_name = osp.splitext(temp_config_name)[0]

GroundingDINO/groundingdino/version.py

@@ -1 +1 @@
-__version__ = '0.1.0'
+__version__ = "0.1.0"