Add python library for yolov7 node

src/detect_ros.py

@@ -17,6 +17,7 @@
 from vision_msgs.msg import Detection2DArray, Detection2D, BoundingBox2D
 from sensor_msgs.msg import Image
 from cv_bridge import CvBridge
+from detection_server import DetectionServer
 import yaml
 
 
@@ -152,6 +153,9 @@ def process_img_msg(self, img_msg: Image):
         # publishing
         detection_msg = create_detection_msg(img_msg, detections)
         self.detection_publisher.publish(detection_msg)
+        # post to detection library
+        DetectionServer.update_detections(np_img_resized, detections)
+
 
         # visualizing if required
         if self.visualization_publisher:

src/detection_server.py

@@ -0,0 +1,82 @@
+import rospy
+import torch
+from numpy import ndarray
+from utils.Detection2DArray import Detection2DArray
+from utils.Detection2D import Detection2D 
+from utils.BoundingBox2D import BoundingBox2D
+from utils.ObjectHypothesisWithPose import ObjectHypothesisWithPose
+
+# todo: update to classes
+class DetectionServer:
+    def __init__(self): 
+        self.detection_array = Detection2DArray()
+        self.detection_history = []
+        self.number_of_detections = 0
+        self.objects_found = False
+
+# todo: find our if I still need this
+    def create_time_stamp():
+        h = Header()
+        h.stamp = rospy.Time.now()
+        return h
+
+
+    def update_detections(image: ndarray, detections: torch.Tensor) -> Detection2DArray:
+        """
+        :param detections: torch tensor of shape [num_boxes, 6] where each element is
+            [x1, y1, x2, y2, confidence, class_id]
+        :returns: detections as a ros message of type Detection2DArray
+        """
+        self.detection_array = Detection2DArray()
+
+        # todo: add source image https://docs.ros.org/en/lunar/api/vision_msgs/html/msg/Detection2D.html
+
+        # header
+        time_stamp = create_time_stamp()
+        self.detection_array.header = time_stamp
+        self.detection_array.img = image
+        for detection in detections:
+            x1, y1, x2, y2, conf, cls = detection.tolist()
+            single_detection_msg = Detection2D()
+            single_detection_msg.header = time_stamp
+
+
+            # bbox
+            bbox = BoundingBox2D()
+            w = int(round(x2 - x1))
+            h = int(round(y2 - y1))
+            cx = int(round(x1 + w / 2))
+            cy = int(round(y1 + h / 2))
+            bbox.size_x = w
+            bbox.size_y = h
+            bbox.center_x = cx
+            bbox.center_y = cy
+
+            single_detection_msg.bbox = bbox
+
+            # class id & confidence
+            obj_hyp = ObjectHypothesisWithPose()
+            obj_hyp.id = int(cls)
+            obj_hyp.score = conf
+            single_detection_msg.results = [obj_hyp]
+
+            self.detection_array.detections.append(single_detection_msg)
+        self.detection_history.append(self.detection_array)
+        return self.detection_array
+
+    def get_detections(): 
+        if(len(self.detection_array)==0:
+           raise Exception("Detections were requested but no objects were found")
+        else: 
+           return self.detection_array
+
+    def get_number_of_objects_detected():
+        return len(self.detection_array)
+
+    def get_detection_history():
+           return self.detection_history
+
+    def erase_detection_history():
+           self.detection_history = []
+
+

src/utils/BoundingBox2D.py

@@ -0,0 +1,7 @@
+class BoundingBox2D: 
+    def __init__(self, cx: int, cy: int, sx: int, sy: int):
+        self.center_x = cx
+        self.center_y = cy
+        self.size_x = sx
+        self.size_y = sy
+

src/utils/Detection2D.py

@@ -0,0 +1,8 @@
+from ObjectHypothesisWithPose import ObjectHypothesisWithPose
+from BoundingBox1D import BoundingBox2D
+
+class Detection2D:
+    def __init__(self, results: ObjectHypothesisWithPose, bbox: BoundingBox2D, header):
+        self.results = results
+        self.bbox = bbox
+        self.header = header

src/utils/Detection2DArray.py

@@ -0,0 +1,7 @@
+from Detection2D import Detection2D
+from numpy import ndarray
+class Detection2DArray: 
+    def __init__(self, img: ndarray, detections: list[Detection2D], header):
+        self.img = img
+        self.detections = detections
+        self.header = header

src/utils/ObjectHypothesisWithPose.py

@@ -0,0 +1,5 @@
+class ObjectHypothesisWithPose: 
+    def __init__(self, id: int, score: float):
+        self.id = id
+        self.score = score
+