Namburger

FROM ubuntu:18.04

WORKDIR /home
ENV HOME /home
RUN cd ~
RUN apt-get update
RUN apt-get install -y git nano python3-pip python-dev pkg-config wget usbutils curl

RUN echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" \
| tee /etc/apt/sources.list.d/coral-edgetpu.list

Namburger

FROM arm64v8/debian:latest

WORKDIR /home
ENV HOME /home
RUN cd ~
RUN apt-get update
RUN apt-get install -y git nano python3-pip python-dev pkg-config wget usbutils curl

RUN echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" \
| tee /etc/apt/sources.list.d/coral-edgetpu.list

Namburger

/dts-v1/;
/plugin/;

/ {
        compatible = "fsl,imx8mq-hdmi";

        fragment@0 {
                target-path = "/hdmi@32c00000";
                __overlay__ {
                        status = "disabled";

Namburger

# Taken from here: https://github.com/googlecreativelab/teachablemachine-community/blob/master/snippets/markdown/image/edgetpu/python.md just fixed some bugs 

from edgetpu.classification.engine import ClassificationEngine
from PIL import Image
import cv2
import re
import os
import sys

# the TFLite converted to be used with edgetpu

Namburger

# Install the edgetpu compiler.
!curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | apt-key add -
!echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
!sudo apt-get update
!sudo apt-get install edgetpu-compiler

# Compile our model and make a tarball of the finished trained model.
%cd /content/output_model
!edgetpu_compiler -s ssdlite_mobiledet_custom.tflite
%cd /content/

Namburger

# Install tflite_runtime package to evaluate the model.
!pip3 install https://dl.google.com/coral/python/tflite_runtime-2.1.0.post1-cp36-cp36m-linux_x86_64.whl

# Now we do evaluation on the tflite model.
import os
import numpy as np
from tflite_runtime.interpreter import Interpreter
from tflite_runtime.interpreter import load_delegate
from PIL import Image
from PIL import ImageDraw

Namburger

# Now we can convert this custom trained model to a CPU tflite model
!tflite_convert \
  --output_file="/content/output_model/ssdlite_mobiledet_dog_vs_cat.tflite" \
  --graph_def_file="/content/output_model/tflite_graph.pb" \
  --inference_type=QUANTIZED_UINT8 \
  --input_arrays="normalized_input_image_tensor" \
  --output_arrays="TFLite_Detection_PostProcess,TFLite_Detection_PostProcess:1,TFLite_Detection_PostProcess:2,TFLite_Detection_PostProcess:3" \
  --mean_values=128 \
  --std_dev_values=128 \
  --input_shapes=1,320,320,3 \

Namburger

# Now we export this model to tflite_graph format.
%cd /content/models/research
!mkdir /content/output_model
!python3 object_detection/export_tflite_ssd_graph.py \
  --pipeline_config_path=/content/models/research/object_detection/samples/configs/ssdlite_mobiledet_edgetpu_320x320_coco_sync_4x4.config \
  --trained_checkpoint_prefix=/content/train/model.ckpt-25000 \
  --output_directory=/content/output_model \
  --add_postprocessing_op=true
# Make sure to change the model-ckpt-# to match the checkpoint number you used.

Namburger

# Do a Quick Evaluation on the inference graph model.
import numpy as np
import os
import sys
import tensorflow as tf

from collections import defaultdict
from matplotlib import pyplot as plt
from PIL import Image

Namburger

# Let's download some test data from flickr.
!mkdir /content/test
!cd /content/test
!wget https://live.staticflickr.com/7921/46683787864_86c9501c24_c_d.jpg -O /content/test/image1.jpg
!wget https://live.staticflickr.com/4/8451898_8bedb2ae53_c_d.jpg -O /content/test/image2.jpg
!wget https://live.staticflickr.com/2654/3997966238_f454845087_c_d.jpg -O /content/test/image3.jpg
!wget https://live.staticflickr.com/2818/34032378096_5309537c9f_c_d.jpg -O /content/test/image4.jpg
!wget https://live.staticflickr.com/8682/28214087384_4c7711584d_c_d.jpg -O /content/test/image5.jpg