feat(perception): define semantic object understanding

2026-08-25 16:43:47 +03:00
parent 1084984da2
commit 6276bbf324
28 changed files with 7028 additions and 0 deletions
@@ -0,0 +1,86 @@
+{
+  "schema_version": "missioncore.open-vocabulary-semantic-shadow-profile/v0",
+  "profile_id": "m48s-raw-kb4-urban-semantics-shadow/v0",
+  "status": "experimental-shadow",
+  "source": {
+    "session_id": "20260720T065719Z_viewer_live",
+    "source_id": "RAVNOVES00",
+    "camera_source_id": "sensor.camera.right",
+    "coordinate_space": "raw-kb4",
+    "width": 800,
+    "height": 600,
+    "decoded_frame_basis": "zero-based-decoded-video-frame",
+    "decoded_image_encoding": "jpeg-q2-yuvj420p",
+    "valid_fov_result_id": "valid-fov-mask-b4dd8ddf2b87c1d520ee8a0868c4fea062d7c14d1bae73ccabd3abe1f3acbac2",
+    "valid_fov_mask_sha256": "a40cee06b7c6f69b6a09a11563dcfd237f3de833b1ccd31459e66692e528ba63",
+    "valid_fov_fill_value": 114
+  },
+  "provider": {
+    "provider_id": "nvidia-tao-grounding-dino-trt/v1",
+    "name": "NVIDIA TAO Grounding DINO Swin-Tiny Commercial",
+    "version": "1.0",
+    "model_id": "grounding-dino-swin-tiny-commercial/v1",
+    "model_file": "grounding_dino_swin_tiny_commercial_deployable.onnx",
+    "model_sha256": "6895acdc6b588e923f753e37b3bd18869e064256e5ecc1b2b9853e8c51125f94",
+    "model_revision": "nvidia-tao-7.0.1-commercial-v1.0",
+    "engine_file": "grounding_dino_swin_tiny_commercial_fp16.engine",
+    "engine_sha256": "63213f60730ce8577047ff508dfe06113269347501e2dd832aeb56c89bc11489",
+    "engine_precision": "FP16",
+    "container_reference": "nvcr.io/nvidia/tao/tao-toolkit:7.0.1-deploy",
+    "container_image_id": "sha256:2a3095330dd83e4314aada21fc7d184fcdb442a9131d591e5593ca5497394e7a",
+    "preprocess_id": "raw-kb4-valid-fov-fill114-800x600-to-960x544/v1",
+    "engine_input_width": 960,
+    "engine_input_height": 544
+  },
+  "postprocessing": {
+    "minimum_input_confidence": 0.25,
+    "minimum_box_area_fraction": 0.00002,
+    "maximum_box_area_fraction": 0.35,
+    "minimum_valid_fov_fraction": 0.5,
+    "require_center_inside_valid_fov": true,
+    "fusion_iou_threshold": 0.7,
+    "selected_minimum_confidence": 0.5,
+    "selected_minimum_margin": 0.1,
+    "max_hypotheses": 5
+  },
+  "prompt_groups": [
+    {
+      "prompt_set_id": "urban-static/v0",
+      "captions": [
+        "trash bin",
+        "shopping cart",
+        "bollard",
+        "post",
+        "concrete hemisphere",
+        "traffic cone",
+        "road sign",
+        "temporary barrier"
+      ]
+    },
+    {
+      "prompt_set_id": "urban-agents/v0",
+      "captions": [
+        "adult person",
+        "child",
+        "dog",
+        "stroller",
+        "kick scooter",
+        "bicycle"
+      ]
+    },
+    {
+      "prompt_set_id": "urban-vehicles/v0",
+      "captions": ["car", "truck", "bus", "motorcycle"]
+    }
+  ],
+  "semantic_vocabulary_id": "missioncore.urban-object-semantics/v0",
+  "risk_policy_id": "urban-object-risk-shadow/v0",
+  "authority": {
+    "ground_truth": false,
+    "independent_truth": false,
+    "candidate_accepted": false,
+    "commands_enabled": false,
+    "actuation_allowed": false,
+    "navigation_or_safety_accepted": false
+  }
+}