rhetenor/awesome-awesomeness
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
Files
652812eed0d0ae17376945242133acb706f9b01f
awesome-awesomeness/html/tensorflowlite.html
rhetenor 5916c5c074 update lists
2025-07-18 22:22:32 +02:00

971 lines
38 KiB
HTML
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
<p align="center">
<img src="images/awesome-tflite.png" alt="awesome tflite" width="500">
</p>
<!-- omit in toc -->
<h1 id="awesome-tensorflow-lite-awesome-prs-welcome-twitter">Awesome
TensorFlow Lite <a href="https://awesome.re"><img
src="https://awesome.re/badge.svg" alt="Awesome" /></a> <a
href="http://makeapullrequest.com"><img
src="https://img.shields.io/badge/PRs-welcome-brightgreen.svg?style=flat-square"
alt="PRs Welcome" /></a> <a href="https://twitter.com/margaretmz"><img
src="https://img.shields.io/badge/Twitter-%40margaretmz-blue"
alt="Twitter" /></a></h1>
<p><a href="https://www.tensorflow.org/lite">TensorFlow Lite</a> is a
set of tools that help convert and optimize TensorFlow models to run on
mobile and edge devices. It’s currently running on more than 4 billion
devices! With TensorFlow 2.x, you can train a model with tf.Keras,
easily convert a model to .tflite and deploy it; or you can download a
pretrained TensorFlow Lite model from the model zoo.</p>
<p>This is an awesome list of TensorFlow Lite models with sample apps,
helpful tools and learning resources - * Showcase what the community has
built with TensorFlow Lite * Put all the samples side-by-side for easy
reference * Share knowledge and learning resources</p>
<p>Please submit a PR if you would like to contribute and follow the
guidelines <a href="CONTRIBUTING.md">here</a>.</p>
<!-- omit in toc -->
<p>## Contents - <a href="#past-announcements">Past announcements:</a> -
<a href="#models-with-samples">Models with samples</a> - <a
href="#computer-vision">Computer vision</a> - <a
href="#classification">Classification</a> - <a
href="#detection">Detection</a> - <a
href="#segmentation">Segmentation</a> - <a href="#style-transfer">Style
Transfer</a> - <a href="#generative">Generative</a> - <a
href="#post-estimation">Post estimation</a> - <a href="#other">Other</a>
- <a href="#text">Text</a> - <a href="#speech">Speech</a> - <a
href="#recommendation">Recommendation</a> - <a href="#game">Game</a> -
<a href="#model-zoo">Model zoo</a> - <a
href="#tensorflow-lite-models">TensorFlow Lite models</a> - <a
href="#tensorflow-models">TensorFlow models</a> - <a
href="#ideas-and-inspiration">Ideas and Inspiration</a> - <a
href="#ml-kit-examples">ML Kit examples</a> - <a
href="#plugins-and-sdks">Plugins and SDKs</a> - <a
href="#helpful-links">Helpful links</a> - <a
href="#learning-resources">Learning resources</a> - <a
href="#blog-posts">Blog posts</a> - <a href="#books">Books</a> - <a
href="#videos">Videos</a> - <a href="#podcasts">Podcasts</a> - <a
href="#moocs">MOOCs</a></p>
<h2 id="past-announcements">Past announcements:</h2>
<p>Here are some past feature annoucements of TensorFlow Lite: * <a
href="https://groups.google.com/a/tensorflow.org/d/msg/tflite/Z_h7706dt8Q/sNrjPj4yGgAJ">Announcement
of the new converter</a> - <a
href="https://medium.com/tensorflow/mlir-a-new-intermediate-representation-and-compiler-framework-beba999ed18d">MLIR</a>-based
and enables conversion of new classes of models such as Mask R-CNN and
Mobile BERT etc., supports functional control flow and better error
handling during conversion. Enabled by default in the nightly builds. *
<a
href="https://github.com/tensorflow/tflite-support/tree/master/tensorflow_lite_support/java">Android
Support Library</a> - Makes mobile development easier (<a
href="https://github.com/tensorflow/examples/blob/master/lite/examples/image_classification/android/EXPLORE_THE_CODE.md">Android</a>
sample code). * <a
href="https://www.tensorflow.org/lite/guide/model_maker">Model Maker</a>
- Create your custom <a
href="https://github.com/tensorflow/examples/tree/master/tensorflow_examples/lite/model_maker">image
&amp; text</a> classification models easily in a few lines of code. See
below the Icon Classifier for a tutorial by the community. * <a
href="https://blog.tensorflow.org/2019/12/example-on-device-model-personalization.html">On-device
training</a> - It is finally here! Currently limited to transfer
learning for image classification only but it’s a great start. See the
official <a
href="https://github.com/tensorflow/examples/blob/master/lite/examples/model_personalization/README.md">Android</a>
sample code and another one from the community (<a
href="https://aqibsaeed.github.io/on-device-activity-recognition">Blog</a>
| <a
href="https://github.com/aqibsaeed/on-device-activity-recognition">Android</a>).
* <a
href="https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/g3doc/performance/hexagon_delegate.md">Hexagon
delegate</a> - How to use the Hexagon Delegate to speed up model
inference on mobile and edge devices. Also see blog post <a
href="https://blog.tensorflow.org/2019/12/accelerating-tensorflow-lite-on-qualcomm.html">Accelerating
TensorFlow Lite on Qualcomm Hexagon DSPs</a>. * <a
href="https://www.tensorflow.org/lite/convert/metadata">Model
Metadata</a> - Provides a standard for model descriptions which also
enables <a
href="https://www.tensorflow.org/lite/inference_with_metadata/codegen">Code
Gen and Android Studio ML Model Binding</a>.</p>
<h2 id="models-with-samples">Models with samples</h2>
<p>Here are the TensorFlow Lite models with app / device
implementations, and references. Note: pretrained TensorFlow Lite models
from MediaPipe are included, which you can implement with or without
MediaPipe.</p>
<h3 id="computer-vision">Computer vision</h3>
<h4 id="classification">Classification</h4>
<table>
<colgroup>
<col style="width: 5%" />
<col style="width: 26%" />
<col style="width: 65%" />
<col style="width: 3%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Classification</td>
<td>MobileNetV1 (<a
href="https://storage.googleapis.com/download.tensorflow.org/models/tflite/mobilenet_v1_1.0_224_quant_and_labels.zip">download</a>)</td>
<td><a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/image_classification/android">Android</a>
| <a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/image_classification/ios">iOS</a>
| <a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/image_classification/raspberry_pi">Raspberry
Pi</a> | <a
href="https://www.tensorflow.org/lite/models/image_classification/overview">Overview</a></td>
<td>tensorflow.org</td>
</tr>
<tr class="even">
<td>Classification</td>
<td>MobileNetV2</td>
<td>Recognize Flowers on Android <a
href="https://codelabs.developers.google.com/codelabs/recognize-flowers-with-tensorflow-on-android/#0">Codelab</a>
| <a
href="https://github.com/tensorflow/examples/tree/master/lite/codelabs/flower_classification/android">Android</a></td>
<td>TensorFlow team</td>
</tr>
<tr class="odd">
<td>Classification</td>
<td>MobileNetV2</td>
<td>Skin Lesion Detection <a
href="https://github.com/AakashKumarNain/skin_cancer_detection/tree/master/demo">Android</a></td>
<td>Community</td>
</tr>
<tr class="even">
<td>Classification</td>
<td>MobileNetV2</td>
<td>American Sign Language Detection | <a
href="https://colab.research.google.com/drive/1xsunX7Qj_XWBZwcZLyjsKBg4RI0DNo2-?usp=sharing">Colab
Notebook</a> | <a
href="https://github.com/sayannath/American-Sign-Language-Detection">Android</a></td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Classification</td>
<td>CNN + Quantisation Aware Training</td>
<td>Stone Paper Scissor Detection <a
href="https://colab.research.google.com/drive/1Wdso2N_76E8Xxniqd4C6T1sV5BuhKN1o?usp=sharing">Colab
Notebook</a> | <a
href="https://github.com/sayannath/American-Sign-Language-Detection">Flutter</a></td>
<td>Community</td>
</tr>
<tr class="even">
<td>Classification</td>
<td>EfficientNet-Lite0 (<a
href="https://github.com/margaretmz/icon-classifier/blob/master/ml-code/icons-50.tflite">download</a>)</td>
<td>Icon Classifier <a
href="https://github.com/margaretmz/icon-classifier">Colab &amp;
Android</a> | <a
href="https://medium.com/swlh/icon-classifier-with-tflite-model-maker-9263c0021f72">tutorial
1</a> | <a
href="https://medium.com/@margaretmz/icon-classifier-android-app-1fc0b727f761">tutorial
2</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h4 id="detection">Detection</h4>
<table>
<colgroup>
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Object detection</td>
<td>Quantized COCO SSD MobileNet v1 (<a
href="https://storage.googleapis.com/download.tensorflow.org/models/tflite/coco_ssd_mobilenet_v1_1.0_quant_2018_06_29.zip">download</a>)</td>
<td><a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/object_detection/android">Android</a>
| <a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/object_detection/ios">iOS</a>
| <a
href="https://www.tensorflow.org/lite/models/object_detection/overview#starter_model">Overview</a></td>
<td>tensorflow.org</td>
</tr>
<tr class="even">
<td>Object detection</td>
<td>YOLO</td>
<td><a
href="https://blog.francium.tech/real-time-object-detection-on-mobile-with-flutter-tensorflow-lite-and-yolo-android-part-a0042c9b62c6">Flutter</a>
| <a href="https://arxiv.org/abs/1506.02640">Paper</a></td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Object detection</td>
<td><a
href="https://tfhub.dev/neso613/lite-model/yolo-v5-tflite/tflite_model/1">YOLOv5</a></td>
<td><a href="https://github.com/neso613/yolo-v5-tflite-model">Yolov5
Inference</a></td>
<td>Community</td>
</tr>
<tr class="even">
<td>Object detection</td>
<td>MobileNetV2 SSD (<a
href="https://github.com/google/mediapipe/tree/master/mediapipe/models/ssdlite_object_detection.tflite">download</a>)</td>
<td><a
href="https://github.com/google/mediapipe/blob/master/mediapipe/models/object_detection_saved_model/README.md">Reference</a></td>
<td>MediaPipe</td>
</tr>
<tr class="odd">
<td>Object detection</td>
<td>MobileDet (<a
href="https://arxiv.org/abs/2004.14525">Paper</a>)</td>
<td><a href="https://sayak.dev/mobiledet-optimization/">Blog post
(includes the TFLite conversion process)</a></td>
<td>MobileDet is from University of Wisconsin-Madison and Google and the
blog post is from the Community</td>
</tr>
<tr class="even">
<td>License Plate detection</td>
<td>SSD MobileNet <a
href="https://github.com/ariG23498/Flutter-License/blob/master/assets/detect.tflite">(download)</a></td>
<td><a
href="https://github.com/ariG23498/Flutter-License">Flutter</a></td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Face detection</td>
<td>BlazeFace (<a
href="https://github.com/google/mediapipe/tree/master/mediapipe/models/face_detection_front.tflite">download</a>)</td>
<td><a
href="https://sites.google.com/corp/view/perception-cv4arvr/blazeface">Paper</a></td>
<td>MediaPipe</td>
</tr>
<tr class="even">
<td>Face Authentication</td>
<td><a href="https://arxiv.org/pdf/1503.03832.pdf">FaceNet</a></td>
<td><a
href="https://github.com/sayannath/Face-Authentication-App">Flutter</a></td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Hand detection &amp; tracking</td>
<td>Palm detection &amp; hand landmarks (<a
href="https://github.com/google/mediapipe/tree/master/mediapipe/models#hand-detection-and-tracking">download</a>)</td>
<td><a href="https://mediapipe.page.link/handgoogleaiblog">Blog post</a>
| <a href="https://mediapipe.page.link/handmc">Model card</a> | <a
href="https://github.com/supremetech/mediapipe-demo-hand-detection">Android</a></td>
<td>MediaPipe &amp; Community</td>
</tr>
</tbody>
</table>
<h4 id="segmentation">Segmentation</h4>
<table>
<colgroup>
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Segmentation</td>
<td>DeepLab V3 (<a
href="https://storage.googleapis.com/download.tensorflow.org/models/tflite/gpu/deeplabv3_257_mv_gpu.tflite">download</a>)</td>
<td><a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/image_segmentation/">Android
&amp; iOS</a> | <a
href="https://www.tensorflow.org/lite/models/segmentation/overview">Overview</a>
| Flutter <a
href="https://github.com/kshitizrimal/Flutter-TFLite-Image-Segmentation">Image</a>
| <a
href="https://github.com/kshitizrimal/tflite-realtime-flutter">Realtime</a>
| <a href="https://arxiv.org/abs/1706.05587">Paper</a></td>
<td>tf.org &amp; Community</td>
</tr>
<tr class="even">
<td>Segmentation</td>
<td>Different variants of <a
href="https://github.com/tensorflow/models/blob/master/research/deeplab/g3doc/model_zoo.md">DeepLab
V3 models</a></td>
<td>Models on <a
href="https://tfhub.dev/s?module-type=image-segmentation&amp;publisher=sayakpaul">TF
Hub</a> with Colab Notebooks</td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Segmentation</td>
<td><a
href="https://tfhub.dev/tensorflow/lite-model/deeplabv3/1/metadata/2?lite-format=tflite">DeepLab
V3 model</a></td>
<td><a
href="https://github.com/farmaker47/Update_image_segmentation">Android</a>
| <a
href="https://farmaker47.medium.com/use-camerax-with-image-segmentation-android-project-d8656f35cea3">Tutorial</a></td>
<td>Community</td>
</tr>
<tr class="even">
<td>Hair Segmentation</td>
<td><a
href="https://github.com/google/mediapipe/tree/master/mediapipe/models/hair_segmentation.tflite">Download</a></td>
<td><a
href="https://sites.google.com/corp/view/perception-cv4arvr/hair-segmentation">Paper</a></td>
<td>MediaPipe</td>
</tr>
</tbody>
</table>
<h4 id="style-transfer">Style Transfer</h4>
<table>
<colgroup>
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Style transfer</td>
<td><a
href="https://github.com/tensorflow/magenta/tree/master/magenta/models/arbitrary_image_stylization">Arbitrary
image stylization</a></td>
<td><a
href="https://www.tensorflow.org/lite/models/style_transfer/overview">Overview</a>
| <a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/style_transfer/android">Android</a>
| <a
href="https://github.com/PuzzleLeaf/flutter_tflite_style_transfer">Flutter</a></td>
<td>tf.org &amp; Community</td>
</tr>
<tr class="even">
<td>Style transfer</td>
<td>Better-quality style transfer models in .tflite</td>
<td>Models on <a
href="https://tfhub.dev/sayakpaul/lite-model/arbitrary-image-stylization-inceptionv3/dr/predict/1">TF
Hub</a> with Colab Notebooks</td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Video Style Transfer</td>
<td>Download: <br> <a
href="https://tfhub.dev/sayakpaul/lite-model/arbitrary-image-stylization-inceptionv3-dynamic-shapes/dr/transfer/1">Dynamic
range models</a>)</td>
<td><a
href="https://github.com/farmaker47/video_style_transfer">Android</a> |
<a
href="https://medium.com/@farmaker47/android-implementation-of-video-style-transfer-with-tensorflow-lite-models-9338a6d2a3ea">Tutorial</a></td>
<td>Community</td>
</tr>
<tr class="even">
<td>Segmentation &amp; Style transfer</td>
<td>DeepLabV3 &amp; Style Transfer <a
href="https://github.com/margaretmz/segmentation-style-transfer/tree/master/ml">models</a></td>
<td><a
href="https://github.com/margaretmz/segmentation-style-transfer">Project
repo</a> | <a
href="https://github.com/margaretmz/segmentation-style-transfer/tree/master/android">Android</a>
| <a
href="https://medium.com/google-developer-experts/image-background-stylizer-part-1-project-intro-d68c4547e7e3">Tutorial</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h4 id="generative">Generative</h4>
<table>
<colgroup>
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>GANs</td>
<td><a href="https://github.com/taki0112/UGATIT">U-GAT-IT</a>
(Selfie2Anime)</td>
<td><a
href="https://github.com/margaretmz/selfie2anime-with-tflite">Project
repo</a> | <a
href="https://github.com/margaretmz/selfie2anime-with-tflite/tree/master/android">Android</a>
| <a
href="https://medium.com/google-developer-experts/selfie2anime-with-tflite-part-1-overview-f97500800ffe">Tutorial</a></td>
<td>Community</td>
</tr>
<tr class="even">
<td>GANs</td>
<td><a
href="https://github.com/SystemErrorWang/White-box-Cartoonization">White-box
CartoonGAN</a> (<a
href="https://tfhub.dev/sayakpaul/lite-model/cartoongan/dr/1">download</a>)</td>
<td><a
href="https://github.com/margaretmz/Cartoonizer-with-TFLite">Project
repo</a> | <a
href="https://github.com/margaretmz/Cartoonizer-with-TFLite/tree/master/android">Android</a>
| <a
href="https://blog.tensorflow.org/2020/09/how-to-create-cartoonizer-with-tf-lite.html">Tutorial</a></td>
<td>Community</td>
</tr>
<tr class="odd">
<td>GANs - Image Extrapolation</td>
<td>Boundless on <a
href="https://tfhub.dev/sayakpaul/lite-model/boundless-quarter/dr/1">TF
Hub</a></td>
<td><a
href="https://colab.research.google.com/github/sayakpaul/Adventures-in-TensorFlow-Lite/blob/master/Boundless_TFLite.ipynb">Colab
Notebook</a> | <a href="https://arxiv.org/pdf/2003.06792v2.pdf">Original
Paper</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h4 id="post-estimation">Post estimation</h4>
<table>
<colgroup>
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Pose estimation</td>
<td>Posenet (<a
href="https://storage.googleapis.com/download.tensorflow.org/models/tflite/posenet_mobilenet_v1_100_257x257_multi_kpt_stripped.tflite">download</a>)</td>
<td><a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/posenet/android">Android</a>
| <a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/posenet/ios">iOS</a>
| <a
href="https://www.tensorflow.org/lite/models/pose_estimation/overview">Overview</a></td>
<td>tensorflow.org</td>
</tr>
<tr class="even">
<td>Pose Classification based Video Game Control</td>
<td>MoveNet Lightning (<a
href="https://github.com/NSTiwari/Video-Game-Control-using-Pose-Classification-and-TensorFlow-Lite/blob/main/movenet_lightning.tflite">download</a>)</td>
<td><a
href="https://github.com/NSTiwari/Video-Game-Control-using-Pose-Classification-and-TensorFlow-Lite">Project
Repository</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h4 id="other">Other</h4>
<table>
<colgroup>
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
<col style="width: 25%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Low-light image enhancement</td>
<td><a href="https://tfhub.dev/sayakpaul/mirnet-fixed/1">Models on TF
Hub</a></td>
<td><a href="https://github.com/sayakpaul/MIRNet-TFLite">Project
repo</a> | <a href="https://arxiv.org/pdf/2003.06792v2.pdf">Original
Paper</a> | <a
href="https://github.com/sayannath/MIRNet-Flutter">Flutter</a></td>
<td></td>
</tr>
<tr class="even">
<td>OCR</td>
<td><a
href="https://tfhub.dev/tulasiram58827/lite-model/keras-ocr/dr/2">Models
on TF Hub</a></td>
<td><a href="https://github.com/tulasiram58827/ocr_tflite">Project
Repository</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h3 id="text">Text</h3>
<table>
<colgroup>
<col style="width: 4%" />
<col style="width: 31%" />
<col style="width: 59%" />
<col style="width: 4%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>Sample apps</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Question &amp; Answer</td>
<td>DistilBERT</td>
<td><a
href="https://github.com/huggingface/tflite-android-transformers/blob/master/bert">Android</a></td>
<td>Hugging Face</td>
</tr>
<tr class="even">
<td>Text Generation</td>
<td>GPT-2 / DistilGPT2</td>
<td><a
href="https://github.com/huggingface/tflite-android-transformers/blob/master/gpt2">Android</a></td>
<td>Hugging Face</td>
</tr>
<tr class="odd">
<td>Text Classification</td>
<td><a
href="https://storage.googleapis.com/download.tensorflow.org/models/tflite/text_classification/text_classification.tflite">Download</a></td>
<td><a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/text_classification/android">Android</a>
|<a href="https://github.com/khurram18/TextClassafication">iOS</a> | <a
href="https://github.com/am15h/tflite_flutter_plugin/tree/master/example">Flutter</a></td>
<td>tf.org &amp; Community</td>
</tr>
<tr class="even">
<td>Text Detection</td>
<td>CRAFT Text Detector (<a
href="https://arxiv.org/pdf/1904.01941">Paper</a>)</td>
<td><a
href="https://github.com/tulasiram58827/craft_tflite/blob/main/models/craft_float_800.tflite?raw=true">Download</a>
| <a href="https://github.com/tulasiram58827/craft_tflite/">Project
Repository</a> | <a
href="https://tulasi.dev/craft-in-tflite">Blog1-Conversion to TFLite</a>
| <a href="https://sayak.dev/optimizing-text-detectors/">Blog2-EAST vs
CRAFT</a> | <a
href="https://tfhub.dev/tulasiram58827/lite-model/craft-text-detector/dr/1">Models
on TF Hub</a> | Android (Coming Soon)</td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Text Detection</td>
<td>EAST Text Detector (<a
href="https://arxiv.org/abs/1704.03155">Paper</a>)</td>
<td><a
href="https://tfhub.dev/sayakpaul/lite-model/east-text-detector/dr/1">Models
on TF Hub</a> | <a
href="https://colab.research.google.com/github/sayakpaul/Adventures-in-TensorFlow-Lite/blob/master/EAST_TFLite.ipynb">Conversion
and Inference Notebook</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h3 id="speech">Speech</h3>
<table>
<colgroup>
<col style="width: 11%" />
<col style="width: 23%" />
<col style="width: 56%" />
<col style="width: 7%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Speech Recognition</td>
<td>DeepSpeech</td>
<td><a
href="https://github.com/mozilla/DeepSpeech/tree/master/native_client/java">Reference</a></td>
<td>Mozilla</td>
</tr>
<tr class="even">
<td>Speech Recognition</td>
<td>CONFORMER</td>
<td><a href="https://github.com/neso613/ASR_TFLite">Inference</a> <a
href="https://github.com/windmaple/tflite-asr">Android</a></td>
<td>Community</td>
</tr>
<tr class="odd">
<td>Speech Synthesis</td>
<td>Tacotron-2, FastSpeech2, MB-Melgan</td>
<td><a
href="https://github.com/TensorSpeech/TensorflowTTS/tree/master/examples/android">Android</a></td>
<td>TensorSpeech</td>
</tr>
<tr class="even">
<td>Speech Synthesis(TTS)</td>
<td>Tacotron2, FastSpeech2, MelGAN, MB-MelGAN, HiFi-GAN, Parallel
WaveGAN</td>
<td><a
href="https://github.com/tulasiram58827/TTS_TFLite/blob/main/End_to_End_TTS.ipynb">Inference
Notebook</a> | <a
href="https://github.com/tulasiram58827/TTS_TFLite/">Project
Repository</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h3 id="recommendation">Recommendation</h3>
<table>
<colgroup>
<col style="width: 11%" />
<col style="width: 23%" />
<col style="width: 56%" />
<col style="width: 7%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>On-device Recommendation</td>
<td><a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/recommendation/ml">Dual-Encoder</a></td>
<td><a
href="https://github.com/tensorflow/examples/tree/master/lite/examples/recommendation/android">Android</a>
| <a
href="https://github.com/zhuzilin/on-device_recommendation_tflite">iOS</a>
| <a
href="https://blog.tensorflow.org/2020/09/introduction-to-tflite-on-device-recommendation.html">Reference</a></td>
<td>tf.org &amp; Community</td>
</tr>
</tbody>
</table>
<h3 id="game">Game</h3>
<table>
<colgroup>
<col style="width: 11%" />
<col style="width: 23%" />
<col style="width: 56%" />
<col style="width: 7%" />
</colgroup>
<thead>
<tr class="header">
<th>Task</th>
<th>Model</th>
<th>App | Reference</th>
<th>Source</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>Game agent</td>
<td>Reinforcement learning</td>
<td><a
href="https://github.com/windmaple/planestrike-flutter">Flutter</a> | <a
href="https://windmaple.medium.com/">Tutorial</a></td>
<td>Community</td>
</tr>
</tbody>
</table>
<h2 id="model-zoo">Model zoo</h2>
<h3 id="tensorflow-lite-models">TensorFlow Lite models</h3>
<p>These are the TensorFlow Lite models that could be implemented in
apps and things: * <a
href="https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet/README.md">MobileNet</a>
- Pretrained MobileNet v2 and v3 models. * TensorFlow Lite models * <a
href="https://www.tensorflow.org/lite/models">TensorFlow Lite models</a>
- With official Android and iOS examples. * <a
href="https://www.tensorflow.org/lite/guide/hosted_models">Pretrained
models</a> - Quantized and floating point variants. * <a
href="https://tfhub.dev/">TensorFlow Hub</a> - Set “Model format =
TFLite” to find TensorFlow Lite models.</p>
<h3 id="tensorflow-models">TensorFlow models</h3>
<p>These are TensorFlow models that could be converted to .tflite and
then implemented in apps and things: * <a
href="https://github.com/tensorflow/models/tree/master/official">TensorFlow
models</a> - Official TensorFlow models. * <a
href="https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf2_detection_zoo.md">Tensorflow
detection model zoo</a> - Pre-trained on COCO, KITTI, AVA v2.1,
iNaturalist Species datasets.</p>
<h2 id="ideas-and-inspiration">Ideas and Inspiration</h2>
<ul>
<li><a href="https://github.com/ml-gde/e2e-tflite-tutorials">E2E TFLite
Tutorials</a> - Checkout this repo for sample app ideas and seeking help
for your tutorial projects. Once a project gets completed, the links of
the TensorFlow Lite model(s), sample code and tutorial will be added to
this awesome list.</li>
</ul>
<h2 id="ml-kit-examples">ML Kit examples</h2>
<p><a href="https://developers.google.com/ml-kit">ML Kit</a> is a mobile
SDK that brings Google’s ML expertise to mobile developers. * 2019-10-01
<a
href="https://codelabs.developers.google.com/codelabs/mlkit-android-translate/#0">ML
Kit Translate demo</a> - A tutorial with material design <a
href="https://github.com/googlecodelabs/mlkit-android/tree/master/translate">Android</a>
(Kotlin) sample - recognize, identify Language and translate text from
live camera with ML Kit for Firebase. * 2019-03-13 <a
href="https://youtu.be/ymyYUCrJnxU">Computer Vision with ML Kit -
Flutter In Focus</a>. * 2019-02-09 <a
href="https://medium.com/flutter-community/flutter-mlkit-8039ec66b6a">Flutter
+ MLKit: Business Card Mail Extractor</a> - A blog post with a <a
href="https://github.com/DaemonLoki/Business-Card-Mail-Extractor">Flutter</a>
sample code. * 2019-02-08 <a
href="https://speakerdeck.com/jinqian/from-tensorflow-to-ml-kit-power-your-android-application-with-machine-learning">From
TensorFlow to ML Kit: Power your Android application with machine
learning</a> - A talk with <a
href="https://github.com/xebia-france/magritte">Android</a> (Kotlin)
sample code. * 2018-08-07 <a
href="https://medium.com/over-engineering/building-a-custom-machine-learning-model-on-android-with-tensorflow-lite-26447e53abf2">Building
a Custom Machine Learning Model on Android with TensorFlow Lite</a>. *
2018-07-20 <a
href="https://flatteredwithflutter.com/ml-kit-and-face-detection-in-flutter/">ML
Kit and Face Detection in Flutter</a>. * 2018-07-27 <a
href="https://medium.com/google-developer-experts/exploring-firebase-mlkit-on-android-landmark-detection-part-four-5e86b8deac3a">ML
Kit on Android 4: Landmark Detection</a>. * 2018-07-28 <a
href="https://medium.com/google-developer-experts/exploring-firebase-mlkit-on-android-barcode-scanning-part-three-cc6f5921a108">ML
Kit on Android 3: Barcode Scanning</a>. * 2018-05-31 <a
href="https://medium.com/google-developer-experts/exploring-firebase-mlkit-on-android-face-detection-part-two-de7e307c52e0">ML
Kit on Android 2: Face Detection</a>. * 2018-05-22 <a
href="https://medium.com/google-developer-experts/exploring-firebase-mlkit-on-android-introducing-mlkit-part-one-98fcfedbeee0">ML
Kit on Android 1: Intro</a>.</p>
<h2 id="plugins-and-sdks">Plugins and SDKs</h2>
<ul>
<li><a href="https://www.edgeimpulse.com/">Edge Impulse</a> - Created by
<a href="https://twitter.com/EdgeImpulse"><span class="citation"
data-cites="EdgeImpulse">@EdgeImpulse</span></a> to help you to train
TensorFlow Lite models for embedded devices in the cloud.</li>
<li><a href="https://github.com/google/mediapipe">MediaPipe</a> - A
cross platform (mobile, desktop and Edge TPUs) AI pipeline by Google AI.
(PM <a href="https://twitter.com/realmgyong">Ming Yong</a>) | <a
href="https://mediapipe.readthedocs.io/en/latest/examples.html">MediaPipe
examples</a>.</li>
<li><a href="https://coral.ai/">Coral Edge TPU</a> - Edge hardware by
Google. <a href="https://coral.ai/examples/">Coral Edge TPU
examples</a>.</li>
<li><a href="https://github.com/am15h/tflite_flutter_plugin/">TensorFlow
Lite Flutter Plugin</a> - Provides a dart API similar to the TensorFlow
Lite Java API for accessing TensorFlow Lite interpreter and performing
inference in flutter apps. <a
href="https://pub.dev/packages/tflite_flutter">tflite_flutter on
pub.dev</a>.</li>
</ul>
<h2 id="helpful-links">Helpful links</h2>
<ul>
<li><a href="https://github.com/lutzroeder/netron">Netron</a> - A tool
for visualizing models.</li>
<li><a href="http://ai-benchmark.com/tests.html">AI benchmark</a> - A
website for benchmarking computer vision models on smartphones.</li>
<li><a
href="https://www.tensorflow.org/lite/performance/measurement">Performance
measurement</a> - How to measure model performance on Android and
iOS.</li>
<li><a
href="https://material.io/collections/machine-learning/patterns-for-machine-learning-powered-features.html">Material
design guidelines for ML</a> - How to design machine learning powered
features. A good example: <a
href="https://github.com/firebase/mlkit-material-android">ML Kit
Showcase App</a>.</li>
<li><a href="https://pair.withgoogle.com/">The People + AI Guide
book</a> - Learn how to design human-centered AI products.</li>
<li><a
href="https://github.com/sayakpaul/Adventures-in-TensorFlow-Lite">Adventures
in TensorFlow Lite</a> - A repository showing non-trivial conversion
processes and general explorations in TensorFlow Lite.</li>
<li><a href="https://github.com/iglaweb/TFProfiler">TFProfiler</a> - An
Android-based app to profile TensorFlow Lite models and measure its
performance on smartphone.</li>
<li><a
href="https://www.tensorflow.org/lite/microcontrollers">TensorFlow Lite
for Microcontrollers</a></li>
<li><a
href="https://github.com/dailystudio/tensorflow-lite-examples-android">TensorFlow
Lite Examples - Android</a> - A repository refactors and rewrites all
the TensorFlow Lite Android examples which are included in the
TensorFlow official website.</li>
<li><a
href="https://github.com/SunitRoy2703/Tensorflow-lite-kotlin-samples">Tensorflow-lite-kotlin-samples</a>
- A collection of Tensorflow Lite Android example Apps in Kotlin, to
show different kinds of kotlin implementation of the <a
href="https://www.tensorflow.org/lite/examples">example apps</a></li>
</ul>
<h2 id="learning-resources">Learning resources</h2>
<p>Interested but not sure how to get started? Here are some learning
resources that will help you whether you are a beginner or a
practitioner in the field for a while.</p>
<h3 id="blog-posts">Blog posts</h3>
<ul>
<li>2021-11-09 <a
href="https://blog.tensorflow.org/2021/11/on-device-training-in-tensorflow-lite.html">On-device
training in TensorFlow Lite</a></li>
<li>2021-09-27 <a
href="https://blog.tensorflow.org/2021/09/blog.tensorflow.org202109optical-character-recognition.html">Optical
character recognition with TensorFlow Lite: A new example app</a></li>
<li>2021-06-16 <a
href="https://blog.tensorflow.org/2021/11/on-device-training-in-tensorflow-lite.html">https://blog.tensorflow.org/2021/06/easier-object-detection-on-mobile-with-tf-lite.html</a></li>
<li>2020-12-29 <a
href="https://medium.com/analytics-vidhya/yolov3-to-tensorflow-lite-conversion-4602cec5c239">YOLOv3
to TensorFlow Lite Conversion</a> - By Nitin Tiwari.</li>
<li>2020-04-20 <a
href="https://blog.tensorflow.org/2020/04/whats-new-in-tensorflow-lite-from-devsummit-2020.html">What
is new in TensorFlow Lite</a> - By Khanh LeViet.</li>
<li>2020-04-17 <a
href="https://blog.tensorflow.org/2020/04/optimizing-style-transfer-to-run-on-mobile-with-tflite.html">Optimizing
style transfer to run on mobile with TFLite</a> - By Khanh LeViet and
Luiz Gustavo Martins.</li>
<li>2020-04-14 <a
href="https://blog.tensorflow.org/2020/04/how-tensorflow-lite-helps-you-from-prototype-to-product.html">How
TensorFlow Lite helps you from prototype to product</a> - By Khanh
LeViet.</li>
<li>2019-11-08 <a
href="https://blog.particle.io/2019/11/08/particle-machine-learning-101/">Getting
Started with ML on MCUs with TensorFlow</a> - By Brandon Satrom.</li>
<li>2019-08-05 <a
href="https://blog.tensorflow.org/2019/08/tensorflow-model-optimization-toolkit_5.html">TensorFlow
Model Optimization Toolkit — float16 quantization halves model size</a>
- By the TensorFlow team.</li>
<li>2018-07-13 <a
href="https://blog.tensorflow.org/2018/07/training-and-serving-realtime-mobile-object-detector-cloud-tpus.html">Training
and serving a real-time mobile object detector in 30 minutes with Cloud
TPUs</a> - By Sara Robinson, Aakanksha Chowdhery, and Jonathan
Huang.</li>
<li>2018-06-11 - <a
href="https://petewarden.com/2018/06/11/why-the-future-of-machine-learning-is-tiny/">Why
the Future of Machine Learning is Tiny</a> - By Pete Warden.</li>
<li>2018-03-30 - <a
href="https://blog.tensorflow.org/2018/03/using-tensorflow-lite-on-android.html">Using
TensorFlow Lite on Android</a>) - By Laurence Moroney.</li>
</ul>
<h3 id="books">Books</h3>
<ul>
<li>2021-12-01 <a
href="https://learning.oreilly.com/library/view/ai-and-machine/9781098101732/">AI
and Machine Learning On-Device Development</a> (early access) - By
Laurence Moroney (<a href="https://twitter.com/lmoroney"><span
class="citation" data-cites="lmoroney">@lmoroney</span></a>).</li>
<li>2020-10-01 <a
href="https://learning.oreilly.com/library/view/ai-and-machine/9781492078180/">AI
and Machine Learning for Coders</a> - By Laurence Moroney (<a
href="https://twitter.com/lmoroney"><span class="citation"
data-cites="lmoroney">@lmoroney</span></a>).</li>
<li>2020-04-06 <a
href="https://www.packtpub.com/product/mobile-deep-learning-with-tensorflow-lite-ml-kit-and-flutter/9781789611212">Mobile
Deep Learning with TensorFlow Lite, ML Kit and Flutter</a>: Build
scalable real-world projects to implement end-to-end neural networks on
Android and iOS (<a
href="https://github.com/PacktPublishing/Mobile-Deep-Learning-Projects">GitHub</a>)
- By Anubhav Singh (<a href="https://github.com/xprilion"><span
class="citation" data-cites="xprilion">@xprilion</span></a>) and Rimjhim
Bhadani (<a href="https://github.com/Rimjhim28"><span class="citation"
data-cites="Rimjhim28">@Rimjhim28</span></a>).</li>
<li>2020-03-01 Raspberry Pi for Computer Vision (<a
href="https://www.pyimagesearch.com/raspberry-pi-for-computer-vision">Complete
Bundle</a> | <a
href="https://www.pyimagesearch.com/2019/04/05/table-of-contents-raspberry-pi-for-computer-vision/">TOC</a>)
- By the PyImageSearch Team: Adrian Rosebrock (<a
href="https://twitter.com/PyImageSearch"><span class="citation"
data-cites="PyImageSearch">@PyImageSearch</span></a>), David Hoffman,
Asbhishek Thanki, Sayak Paul (<a
href="https://twitter.com/RisingSayak"><span class="citation"
data-cites="RisingSayak">@RisingSayak</span></a>), and David
Mcduffee.</li>
<li>2019-12-01 <a
href="http://shop.oreilly.com/product/0636920254508.do">TinyML</a> - By
Pete Warden (<a href="https://twitter.com/petewarden"><span
class="citation" data-cites="petewarden">@petewarden</span></a>) and
Daniel Situnayake (<a href="https://twitter.com/dansitu"><span
class="citation" data-cites="dansitu">@dansitu</span></a>).</li>
<li>2019-10-01 <a href="https://www.practicaldeeplearning.ai/">Practical
Deep Learning for Cloud, Mobile, and Edge</a> - By Anirudh Koul (<a
href="https://twitter.com/AnirudhKoul"><span class="citation"
data-cites="AnirudhKoul">@AnirudhKoul</span></a>), Siddha Ganju (<a
href="https://twitter.com/SiddhaGanju"><span class="citation"
data-cites="SiddhaGanju">@SiddhaGanju</span></a>), and Meher Kasam (<a
href="https://twitter.com/MeherKasam"><span class="citation"
data-cites="MeherKasam">@MeherKasam</span></a>).</li>
</ul>
<h3 id="videos">Videos</h3>
<ul>
<li>2021-10-06 <a href="https://youtu.be/sZayUoWW6nE">Contributing to
TensorFlow Lite with Sunit Roy</a> (Hacktoberfest 2021)</li>
<li>2020-07-25 <a href="https://youtu.be/m_bEh8YifnQ">Android ML by Hoi
Lam</a> (GDG Kolkata meetup).</li>
<li>2020-04-01 <a href="https://youtu.be/ALxWJoh_BHw">Easy on-device ML
from prototype to production</a> (TF Dev Summit 2020).</li>
<li>2020-03-11 <a href="https://youtu.be/27Zx-4GOQA8">TensorFlow Lite:
ML for mobile and IoT devices</a> (TF Dev Summit 2020).</li>
<li>2019-10-31 <a href="https://youtu.be/zjDGAiLqGk8">Keynote -
TensorFlow Lite: ML for mobile and IoT devices</a>.</li>
<li>2019-10-31 <a href="https://youtu.be/0SpZy7iouFU">TensorFlow Lite:
Solution for running ML on-device</a>.</li>
<li>2019-10-31 <a href="https://youtu.be/3JWRVx1OKQQ">TensorFlow model
optimization: Quantization and pruning</a>.</li>
<li>2019-10-29 <a href="https://youtu.be/gHN0jDbJz8E">Inside TensorFlow:
TensorFlow Lite</a>.</li>
<li>2018-04-18 <a href="https://youtu.be/JnhW5tQ_7Vo">TensorFlow Lite
for Android (Coding TensorFlow)</a>.</li>
</ul>
<h3 id="podcasts">Podcasts</h3>
<ul>
<li>2020-08-08 <a
href="https://anchor.fm/talkingwithapples/episodes/Talking-Machine-Learning-with-Hoi-Lam-eiaj7v">Talking
Machine Learning with Hoi Lam</a>.</li>
</ul>
<h3 id="moocs">MOOCs</h3>
<ul>
<li><a
href="https://www.udacity.com/course/intro-to-tensorflow-lite--ud190">Introduction
to TensorFlow Lite</a> - Udacity course by Daniel Situnayake (<span
class="citation" data-cites="dansitu">@dansitu</span>), Paige Bailey (<a
href="https://twitter.com/DynamicWebPaige"><span class="citation"
data-cites="DynamicWebPaige">@DynamicWebPaige</span></a>), and Juan
Delgado.</li>
<li><a
href="https://www.coursera.org/learn/device-based-models-tensorflow">Device-based
Models with TensorFlow Lite</a> - Coursera course by Laurence Moroney
(<a href="https://twitter.com/lmoroney"><span class="citation"
data-cites="lmoroney">@lmoroney</span></a>).</li>
<li><a
href="https://www.edx.org/professional-certificate/harvardx-tiny-machine-learning">The
Future of ML is Tiny and Bright</a> - A series of edX courses created by
Harvard in collaboration with Google. Instructors - Vijay Janapa Reddi,
Laurence Moroney, and Pete Warden.</li>
</ul>
<p><a
href="https://github.com/margaretmz/awesome-tensorflow-lite">tensorflowlite.md
Github</a></p>
Reference in New Issue View Git Blame Copy Permalink