Node-Red: TensorFlow to recognize object and ESP8266 device to show result in Dot Matrix

[Music] hi welcome back to my channel in this video i will show you how to use tensorflow node in node-red to recognize object let's get started [Music] first we must install tensorflow node go to manage palette in install tab find the tensorflow node and click install [Music] after install tensorflow node next install browser utils for injecting image file to tensorflow node find an install tab using keyword contrib browser util then click install [Music] okay next we start to create a simple node first select file inject node to select an image then find a tensorflow node in analysis category to recognize the object [Music] connect the nodes and then find debug node and connect to previous node so we can know payload output from tensorflow node don't forget to deploy the codes now we can test the code click file inject node and select the image file for testing object recognition tensorflow will analyze and show the result to output payload [Music] from the testing image tensorflow recognized the object from image file as airplane and payload output [Music] next we test other image files [Music] so [Music] in the test we can see that tensorflow can recognize objects from image files next we will try to use input from the camera in addition we install image output node to see the camera capture [Music] now we start to create code which the input is from the camera and output camera connect to tensorflow node and then go to debug node to c output payload [Music] we also add image preview node to c camera capture [Music] connect usb camera webcam to computer then point camera to an object and click camera node to start to capture tensorflow will analyze image buffer from camera node and then send the result to output payload [Music] [Applause] [Music] in the next experiment we will add an esp device to display the results on the lead matrix here's the concept we have a laptop with no dread dashboard and a webcam connected [Music] then we have esp device with a button as trigger and connect to a dot matrix to display recognition result [Music] and we use mqtt protocol as communication between esp device and node red if button pressed esp will publish a trigger message to mqtt broker then nodered will receive the trigger message and capture the camera and then start to analyze the captured image the recognition result will be sent to mqtt broker as message esp will receive the message and show the dot matrix [Music] before creating the code we must install webcam node to capture the image this node is different from previous camera node this node has an input as capture command [Music] this is the node-red code starting from mqtt node to receive trigger message from esp device select the broker and topic [Music] then switch node will filter the message if messages trigger payload will proceed to next node change node will send capture command to webcam node to capture the image capture command is set true value to capture payload [Music] this is webcam node configuration output payload from tensorflow will be displayed on text node and sent to esp device via mqtt broker [Music] this is the node-red dashboard i put camera output on the dashboard and display camera capture automatically and there is a text to show the result [Music] this is the actual condition there is a webcam connected to laptop and laptop runs no red dashboard this is esp device connected to micro button and dot matrix to show the recognition result [Music] okay we try the experiment put the object on white paper and push the button to start capture image wait until dot matrix shows the object recognition result [Music] so [Music] [Music] so [Music] [Applause] [Music] so [Music] thank you for watching don't forget to like and subscribe see you on next video

Channel: Yaser Ali Husen

Views: 10,591

Keywords:

Id: eckxkzHyeR4

Length: 9min 11sec (551 seconds)

Published: Sun Jul 03 2022