Brainy Pi

Available to select audience (currently beta)

BrainyPi

TFLite examples

Description

TensorFlow Lite (TFLite) is a collection of tools to convert and optimize TensorFlow models to run on edge devices like Brainy Pi. Tflite allows users to create and train custom models therefore creating boundless applications in the realm of ML and AI.

INFO
This documentation is for Rbian OS version: 0.7.2-beta & for TFlite version 2.11.0.
To check the version of Rbian run the command in terminal
				
					os-version
Note: If the command fails or gives error then Rbian version is < 0.7.2-beta.
To check the version of TFlite run the command in python
				
					import tflite_runtime as tflite
print(tflite.__version__)
Note: All the TFlite models that are used require Tensorflow version 2.

Image Segmentation

  • Image segmentation in computer vision is a task that separates a digital image into multiple parts.We will be implementing image segmentation application on BrainyPi using DeepLab (mobilenetv2_dm05_coco_voc_trainval) trained on PASCAL VOC dataset.
  • 20 Objects List:
    • Person: person
    • Animal: bird, cat, cow, dog, horse, sheep
    • Vehicle: aeroplane, bicycle, boat, bus, car, motorbike, train
    • Indoor: bottle, chair, dining table, potted plant, sofa, tv/monitor
				
					git clone https://github.com/brainypi/BrainyPi-AI-Examples.git
cd BrainyPi-AI-Examples/TFLite/ImageSegmentation
  • Run Image segmentation example
				
					python3 imageSegmentation.py
  • Input
    • Parameter1: –image_dir: Image file location. (default=’images/car.jpg’)
    • Parameter2: –save_dir: Directory path to save the result image. (default=’results/result.jpg’)
  • Output
    • Shows the segmentation and overlay with class labels in the output image.

Object Detection

  • Object detection is a computer vision technique for locating instances of objects in images or videos. We will be implementing object detection application on BrainyPi using starter model trained on COCO 2017 dataset.
  • 91 Objects List
				
					person, bicycle, car, motorcycle, airplane, bus, train, truck, boat, traffic light,
fire hydrant, street sign, stop sign, parking meter, bench, bird, cat, dog, horse,
sheep, cow, elephant, bear, zebra, giraffe, hat, backpack, umbrella, shoe, eye glasses,
handbag, tie, suitcase, frisbee, skis, snowboard, sports ball, kite, baseball bat, baseball glove, 
skateboard, surfboard, tennis racket, bottle, plate, wine glass, cup, fork, knife, spoon, bowl, banana,
apple, sandwich, orange, broccoli, carrot, hot dog, pizza, donut, cake, chair, couch, potted plant, bed,
mirror, dining table, window, desk, toilet, door, tv, laptop, mouse, remote, keyboard, cell phone,
microwave, oven, toaster, sink, refrigerator, blender, book, clock, vase, scissors, teddy bear, hair drier,
toothbrush, hair brush
				
			
				
					git clone https://github.com/brainypi/BrainyPi-AI-Examples.git
cd BrainyPi-AI-Examples/TFLite/ObjectDetection
  • Install prerequisites.sh
				
					bash install-prerequisites.sh
  • Run Object Detection example
				
					python3 objectDetection.py
  • Input
    • Parameter 1: ‘–model’, help=’Provide the path to the TFLite file, default is models/model.tflite’ (default=’models/model.tflite’)
    • Parameter 2: ‘–labels’, help=’Provide the path to the Labels, default is models/labels.txt’ (default=’models/labels.txt’)
    • Parameter 3: ‘–image_dir’, help=’Name of the single image to perform detection on’ (default=’images/test1.jpg’)
    • Parameter 4: ‘–threshold’, help=’Minimum confidence threshold for displaying detected objects’ (default=0.5)
    • Parameter 5: ‘–save_dir’, help=’Directory path to save the image’ (default=’results/result.jpg’)
  • Output
    • Shows predicted object bounding boxes with confidence scores.

Pose Estimation

  • Pose estimation is a computer vision technique that predicts and tracks the location of a person or object. We will be implementing pose estimation application on BrainyPi using MoveNet (movenet/singlepose/lightning/tflite/int8).
  • Prerequisites
  • Clone the repository
				
					git clone https://github.com/brainypi/BrainyPi-AI-Examples.git
cd BrainyPi-AI-Examples/TFLite/PoseEstimation
  • Run Pose Estimation example
				
					python3 poseEstimation.py
  • Input
    • Parameter1: –image_dir: Image file location. (default=’images/pose1.jpeg’)
    • Parameter2: –save_dir: Directory path to save the result image. (default=’results/result.jpg’)
  • Output
    • Marks the keypoints on the image (The model is trained on images with a single person only)

Style Transfer

  • Neural style transfer refers to a optimization technique that manipulate digital images to adopt the appearance or visual style of another image. We will be implementing style transfer application on BrainyPi using magenta/arbitrary-image-stylization-v1-256 from tensorflow hub.
  • Prerequisites
  • Clone the repository
				
					git clone https://github.com/brainypi/BrainyPi-AI-Examples.git
cd BrainyPi-AI-Examples/TFLite/StyleTransfer
  • Run Style Transfer example
				
					python3 styleTransfer.py
  • Input
    • Parameter1: –style_image_dir: style Image file location. (default=’images/style.jpg’)
    • Parameter2: –content_image_dir: content Image file location. (default=’images/content.jpg’)
    • Parameter3: –save_dir: Directory path to save the result image. (default=’results/result.jpeg’)
  • Output
    • Shows the Stylized image as a combination of content and style images.

Super Resolution

  • Super-resolution imaging is a class of techniques that enhance the resolution of an imaging system. We will be implementing Low light image enhancement application on BrainyPi using zero-dce model.
  • Prerequisites
  • Clone the repository
				
					git clone https://github.com/brainypi/BrainyPi-AI-Examples.git
cd BrainyPi-AI-Examples/TFLite/SuperResolution
  • Run Super Resolution example
				
					python3 superResolution.py
  • Input
    • Parameter1: –image_dir: Image file location. (default=’images/low.jpg’)
    • Parameter2: –save_dir: Directory path to save the result image. (default=’results/result.jpg’)
  • Output
    • Saves an enhanced image in the specified directory.
NEED SUPPORT?
First, Ensure version of OS installed and the version this document is intended for match. If they match and yet problem persists. Please use this Forum link for community help.
If you are an enterprise customer please use the ticketing system login provided to you for priority support.

Previous

<< Datasheets

Next

GPIO Examples >>

Index