image processing node tweaks

docker/yolo/Dockerfile

@@ -12,23 +12,23 @@ RUN curl -s https://raw.githubusercontent.com/ros/rosdistro/master/ros.asc | apt
 RUN apt-get update && apt-get install -y ros-melodic-ros-base python3-catkin-tools ros-melodic-catkin
 
 # copy over entrypoint
-COPY ./ros_entrypoint.sh ros_entrypoint.sh
+COPY ./ros_entrypoint.sh /ros_entrypoint.sh
 
 # opencv install
 RUN apt-get update && \
 apt-get install -y opencv-data libopencv-dev
 
 # prepare catkin_work space
-RUN mkdir -p catkin_ws/src
+RUN mkdir -p catkin_ws/src && mkdir -p /tmp/darknetMemories
 # darknet ros package install
 COPY ./yolov4-for-darknet_ros /catkin_ws/src/darknet_package
 # build the enclosed darknet installation
-# RUN cd /catkin_ws/src/darknet_ros/darknet && make
 
 # install stuff for building ros packages
 RUN apt-get install -y python-rosdep python-rosinstall python-rosinstall-generator python-wstool build-essential ros-melodic-cv-bridge ros-melodic-image-transport && rosdep init && rosdep update
 
-# run catkin_make
+# install random pip thing and run catkin_make
+RUN apt-get install -y python-pip && pip install empy
 RUN /bin/bash -c '. /opt/ros/melodic/setup.bash && cd catkin_ws && catkin_make'
 RUN cd /catkin_ws/src/darknet_package/darknet_ros/darknet && make
 

docker/yolo/yolov4-for-darknet_ros/darknet_ros/darknet_ros/config/ros.yaml

 subscribers:
 
   camera_reading:
-    topic: /webcam/image_raw
+    topic: /dji_sdk/main_camera_images
     queue_size: 1
 
 actions:

docker/yolo/yolov4-for-darknet_ros/darknet_ros/darknet_ros/launch/yolo_v4_tiny.launch

@@ -6,6 +6,8 @@
   <arg name="network_param_file" default="$(find darknet_ros)/config/yolov4-tiny.yaml"/>
   <arg name="image" default="webcam/image_raw" />
   
+  <!-- Start image processing node -->
+  <node name="darknetImageProcessor" type="darknetImageProc.py" pkg="darknet_ros"/>
 
   <!-- Include main launch file -->
   <include file="$(find darknet_ros)/launch/darknet_ros.launch">

docker/yolo/yolov4-for-darknet_ros/darknet_ros/darknet_ros/scripts/darknetImageProc.py

-#! /usr/bin/python3
+#! /usr/bin/python
 
 import rospy, sys, math, cv2, numpy as np, std_msgs.msg
 from sensor_msgs.msg import Image
-from darknet_ros_msgs.msg import BoundingBox
+from darknet_ros_msgs.msg import BoundingBoxes
 from std_msgs.msg import Header
 import os.path, getpass
 from cv_bridge import CvBridge, CvBridgeError
@@ -11,59 +11,46 @@ from cv_bridge import CvBridge, CvBridgeError
 class ImageProcessor():
     def __init__(self):
         self.bridge = CvBridge()
-        self.imageSubscriber = rospy.Subscriber("webcam/image_raw", Image, self.imageCallback)
-        self.boxSubscriber = rospy.Subscriber("darknet_ros/bounding_boxes", BoundingBox, self.boxCallback)
+        self.imageSubscriber = rospy.Subscriber("/darknet_ros/detection_image", Image, self.imageCallback, queue_size=1)
+        self.boxSubscriber = rospy.Subscriber("/darknet_ros/bounding_boxes", BoundingBoxes, self.boxCallback, queue_size=1)
         self.params = None
         self.image = None
         self.seq = None
+        self.newImageFlag = False
 
         # stuff that needs to be parameters
         self.objectsOfInterest = ["person", "backpack", "car"] # things we report
-        self.reportConfidence = 0.80 # confidence required for report
-        self.reportFrames = 10 # number of frames at high confidence before reporting
+        self.reportConfidence = 0.90 # confidence required for report
         self.saveDirectory = "/tmp/darknetMemories/"
 
     def boxCallback(self, box):
         # check confidence levels and save image when sure of object
+        boxList = box.bounding_boxes
+    	saveFlag = False
+        highPropObjects = ""
+        for box in boxList:
+            if (box.Class in self.objectsOfInterest and box.probability > self.reportConfidence and np.mod(self.seq, 5) == 0):
+                highPropObjects = highPropObjects + box.Class + "+"
                 saveFlag = True
-        if saveFlag:
+
+        if saveFlag and self.newImageFlag:
             if os.path.isdir(self.saveDirectory):
-                cv2.imwrite(self.saveDirectory + "camera_image_{seq}.jpeg".format(seq = self.seq), self.image)
+                cv2.imwrite(self.saveDirectory + "camera_image_{seq}_{obj}.jpeg".format(seq = self.seq, obj = highPropObjects), self.image)
                 rospy.loginfo("saving image to folder: " + self.saveDirectory + "-> camera_image_{seq}.jpeg".format(seq = self.seq))
-                # cv2.waitKey(10)
+                self.newImageFlag = False
             else:
                 rospy.loginfo("image save process failed!")
 
     def imageCallback(self, frame):
+        self.newImageFlag = True
         # load image from topic for later use
-        try:
         self.image = self.bridge.imgmsg_to_cv2(frame, "bgr8")
         self.seq = frame.header.seq
-        except CvBridgeError as err:
-            rospy.loginfo("there was error!")
-            rospy.loginfo(err)
 
 def main():
     rospy.init_node("darknetImageProcessor")
     iv = ImageProcessor()
-
     rospy.spin()
-    # while not rospy.is_shutdown():
-        # if iv.image is not None:
-        #     if np.mod(iv.seq, 32) == 0:
-        #         directory = "/home/" + getpass.getuser() + "/" + folderName
-        #         if os.path.isdir(directory):
-        #             if windowFlag:
-        #                 cv2.imshow("image window", iv.image)
-        #                 rospy.loginfo("publishing image to window seq: " + iv.seq)
-        #             if saveFlag:
-        #                 cv2.imwrite(directory + "/camera_image_{seq}.jpeg".format(seq = iv.seq), iv.image)
-        #                 rospy.loginfo("saving image to folder: " + directory + "-> camera_image_{seq}.jpeg".format(seq = iv.seq) + " seq: " + str(iv.seq))
-        #             cv2.waitKey(10)
-        #         else:
-        #             rospy.loginfo("camera dir missing!")
-
-    # cv2.destroyAllWindows()
 
 if __name__ == "__main__":
     main()