Add SOT target control and state telemetry with image dimensions

protobuf_definitions/control.proto

@@ -247,3 +247,17 @@ message StartLogStreamingCtrl {
 // Message sent to stop streaming log entries to the client.
 message StopLogStreamingCtrl {
 }
+
+// Issue a command to set the single-object tracking (SOT) target.
+//
+// Send a bounding box to start tracking a new target. The CV pipeline will
+// initialize the SOT tracker on the specified region of the video frame.
+message SetSotTargetCtrl {
+  BoundingBox bounding_box = 1; // Target bounding box.
+  uint32 image_width = 2; // Width of the frame the bounding box was drawn on.
+  uint32 image_height = 3; // Height of the frame the bounding box was drawn on.
+}
+
+// Issue a command to clear the single-object tracking (SOT) target (stop tracking).
+message ClearSotTargetCtrl {
+}

protobuf_definitions/message_formats.proto

@@ -757,6 +757,21 @@ enum PressureSensorType {
 
 // Information about the drone.
 //
+// Information about a loaded computer vision model.
+message CvModelInfo {
+  enum ModelType {
+    MODEL_TYPE_UNSPECIFIED = 0;
+    MODEL_TYPE_DETECTION = 1;
+    MODEL_TYPE_SOT = 2;
+  }
+  string name = 1; // Human-readable model name (from model_meta.json).
+  ModelType type = 2; // Type of CV model.
+  bool running = 3; // Whether the model is currently running.
+  string device = 4; // Execution provider (e.g. "cuda", "tensorrt").
+  string package_id = 5; // Package directory name (e.g. "tinyyolov2_package"). Stable identifier.
+  repeated string labels = 6; // Class names the model can detect (indexed by class_id).
+}
+
 // This message contains serial numbers and version information for
 // internal components in the drone. Primarily used for diagnostics, or to
 // determine the origin of a logfile.
@@ -772,6 +787,7 @@ message DroneInfo {
   bytes bb_uid = 8; // Backbone unique identifier.
   GuestPortInfo gp = 9; // Guest port information.
   PressureSensorType depth_sensor = 11; // Type of depth sensor that is connected to the drone.
+  repeated CvModelInfo cv_models = 12; // List of loaded computer vision models.
 }
 
 // Known error states for the drone.
@@ -1430,13 +1446,26 @@ message BoundingBox {
   uint32 height = 4; // Height of the bounding box (px).
 }
 
+// RLE-encoded binary segmentation mask relative to the bounding box.
+//
+// The mask bitmap has dimensions (mask_width x mask_height) covering the
+// detection's bounding box area. The counts field stores run-length encoded
+// data as packed uint16 little-endian: alternating background/foreground
+// pixel runs starting with background.
+message SegmentationMask {
+  uint32 mask_width = 1;  // Width of the RLE bitmap.
+  uint32 mask_height = 2; // Height of the RLE bitmap.
+  bytes counts = 3;       // RLE counts as packed uint16 little-endian.
+}
+
 // A single object detection from a computer vision model.
 message ObjectDetection {
   BoundingBox bounding_box = 1; // Bounding box of the detected object.
   float confidence = 2; // Detection confidence score (0..1).
   uint32 class_id = 3; // Numeric class identifier from the model.
   string class_name = 4; // Human-readable class name.
   uint32 tracking_id = 5; // Unique ID for tracking the same object across frames.
+  SegmentationMask mask = 6; // Instance segmentation mask (absent if model has no segmentation).
 }
 
 // A list of object detections from a single model for a single video frame.
@@ -1471,3 +1500,18 @@ message OperatorInfo {
   string name = 1; // Full name of the operator.
   string email = 2; // E-mail address of the operator.
 }
+
+// Single-object tracking (SOT) state reported by the computer vision pipeline.
+message SotState {
+  // Current state of the SOT tracker.
+  enum State {
+    STATE_UNSPECIFIED = 0; // Unspecified state.
+    STATE_IDLE = 1; // No target selected, waiting for a target bounding box.
+    STATE_TRACKING = 2; // Actively tracking a target.
+    STATE_LOST = 3; // Target was lost (tracker failed to follow the object).
+  }
+  State state = 1; // Current tracking state.
+  BoundingBox bounding_box = 2; // Current tracked bounding box (valid when TRACKING).
+  uint32 image_width = 3; // Width of the source frame in pixels.
+  uint32 image_height = 4; // Height of the source frame in pixels.
+}

protobuf_definitions/telemetry.proto

@@ -341,3 +341,11 @@ message TurbidityFilterTel {
 message CameraPanTiltZoomTel {
   CameraPanTiltZoom camera_pan_tilt_zoom = 1; // Current pan, tilt, and zoom state.
 }
+
+// Single-object tracking (SOT) state telemetry from the computer vision pipeline.
+//
+// Reports the current tracking state and the tracked bounding box when active.
+// Published periodically while the SOT pipeline is running.
+message SotStateTel {
+  SotState sot_state = 1; // Current SOT state and bounding box.
+}