Add images and overview pages

Author: margaretmz

images/screenshots/classify_sound.gif

@@ -0,0 +1,17 @@
+# Classify pose
+
+Here are some ideas:
+* [Human pose](https://github.com/dronefreak/human-action-classification) - 
+use TF Lite’s [PoseNet](https://www.tensorflow.org/lite/models/pose_estimation/overview) and add a layer of fully connected
+* [Hand pose](https://github.com/Prasad9/Classify-HandGesturePose) - 
+use TF Lite’s [hand model](https://google.github.io/mediapipe/solutions/hands) from MediaPipe
+
+Help needed:
+* Generate ideas
+* Write sample apps and tutorials
+
+<p align="center">
+ <img src="../images/screenshots/pose1.png" width='680'/>
+
+ <img src="../images/screenshots/pose2.png" width='680'/>
+</p>

images/screenshots/pose1.png

@@ -0,0 +1,9 @@
+# Sound classification
+
+Teachable Machine (TM) offers a no-code [solution](https://medium.com/@warronbebster/teachable-machine-tutorial-snap-clap-whistle-4212fd7f3555) to train a sound classifier and exporting the model to TensorFlow Lite.
+
+Current status: blocked by FFT support on TF Lite. Waiting for TF team to resolve.
+
+<p align="center">
+ <img src="../images/screenshots/classify_sound.gif" width='680'/>
+</p>

images/screenshots/pose2.png

@@ -0,0 +1,11 @@
+# Speech commands
+
+There are already sample apps for Android and iOS for speech commands - recognition of speech Commands on mobile, highlighting the spoken word; however, there is no end-to-end tutorial on how to use it.
+
+Help needed:
+* Convert the TF1 training code to TF 2
+* Write e2e tutorial on how to use and customize the sample
+
+<p align="center">
+ <img src="../images/screenshots/speech_command.png" width='680'/>
+</p>

images/screenshots/speech_command.png

@@ -0,0 +1,11 @@
+# SPICE (Pitch Detection)
+
+Here are the Blog post and demo.
+
+Model provided FreddieMeter was just published in TF Lite format.
+
+Let’s build a fun demo app with this model!
+
+<p align="center">
+ <img src="../images/screenshots/spice.png" width='680'/>
+</p>

images/screenshots/spice.png

@@ -0,0 +1,12 @@
+# YOLO
+
+Object detection is one of the most popular on-device ML use cases.
+Help needed:
+
+* Convert an open-source YOLO TF implementation on GitHub to TF Lite.
+* Publish a pre-trained model on TF Hub.
+* Write a tutorial on how to train the model on custom dataset using transfer learning.
+
+<p align="center">
+ <img src="../images/screenshots/yolo.png"/>
+</p>