| Up to higher level directory | |||
| Name | Date | Size | |
|---|---|---|---|
| camera/ | 22-Oct-2020 | ||
| label_image/ | 22-Oct-2020 | ||
| README.md | 22-Oct-2020 | 3.4K | |
README.md
1 # TensorFlow Raspberry Pi Examples 2 3 This folder contains examples of how to build applications for the Raspberry Pi using TensorFlow. 4 5 ## Building the Examples 6 7 - Follow the Raspberry Pi section of the instructions at [tensorflow/contrib/makefile](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/makefile) to compile a static library containing the core TensorFlow code. 8 9 - Install libjpeg, so we can load image files: 10 11 ``` 12 sudo apt-get install -y libjpeg-dev 13 ``` 14 15 - To download the example model you'll need, run these commands: 16 17 ```bash 18 curl https://storage.googleapis.com/download.tensorflow.org/models/inception_dec_2015_stripped.zip \ 19 -o /tmp/inception_dec_2015_stripped.zip 20 unzip /tmp/inception_dec_2015_stripped.zip \ 21 -d tensorflow/contrib/pi_examples/label_image/data/ 22 ``` 23 24 - From the root of the TensorFlow source tree, run `make -f tensorflow/contrib/pi_examples/label_image/Makefile` to build a basic example. 25 26 ## Usage 27 28 Run `tensorflow/contrib/pi_examples/label_image/gen/bin/label_image` to try out image labeling with the default Grace Hopper image. You should several lines of output, with "Military Uniform" shown as the top result, something like this: 29 30 ```bash 31 I tensorflow/contrib/pi_examples/label_image/label_image.cc:384] Running model succeeded! 32 I tensorflow/contrib/pi_examples/label_image/label_image.cc:284] military uniform (866): 0.624293 33 I tensorflow/contrib/pi_examples/label_image/label_image.cc:284] suit (794): 0.0473981 34 I tensorflow/contrib/pi_examples/label_image/label_image.cc:284] academic gown (896): 0.0280926 35 I tensorflow/contrib/pi_examples/label_image/label_image.cc:284] bolo tie (940): 0.0156956 36 I tensorflow/contrib/pi_examples/label_image/label_image.cc:284] bearskin (849): 0.0143348 37 ``` 38 39 Once you've verified that is working, you can supply your own images with `--image=your_image.jpg`, or even with graphs you've trained yourself with the TensorFlow for Poets tutorial using `--graph=your_graph.pb --input=Mul:0 --output=final_result:0`. 40 41 ## Camera Example 42 43 Once you have the simple example running, you can try out a more complex version that 44 reads frames from a camera attached to the Pi. You'll need to install and set up your 45 camera module first. The example uses Video4Linux, so you'll need to install that first. 46 Here's some commands I found necessary to get that set up, and I found more information 47 at this blog post: http://www.richardmudhar.com/blog/2015/02/raspberry-pi-camera-and-motion-out-of-the-box-sparrowcam/ 48 49 ``` 50 sudo bash -c "echo 'bcm2835-v4l2' >> /etc/modules" 51 sudo apt-get install libv4l-dev 52 ``` 53 54 Once that's working, run the following commands to build and run the camera example: 55 56 ```bash 57 make -f tensorflow/contrib/pi_examples/camera/Makefile 58 tensorflow/contrib/pi_examples/camera/gen/bin/camera 59 ``` 60 61 You should see it looping over camera frames as they come in, and printing the top labels 62 to the command line. This is a great starting point for all sorts of fun image recognition 63 applications, especially when you combine it with a custom model you've built using 64 something like the TensorFlow for Poets tutorial. 65 66 The example is designed to work with the Flite speech synthesis tool, so that your Pi 67 can speak any labels that have a high enough score. To enable this, just install the 68 Flite package and then pipe the output of the binary you've built, like this: 69 70 ``` 71 sudo apt-get install flite 72 tensorflow/contrib/pi_examples/camera/gen/bin/camera | xargs -n 1 flite -t 73 ``` 74
