Updated 2018 (markdown)

2018.md

@@ -25,6 +25,49 @@
 
 # Notes
 
+<pre>
+<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
+</pre>
+
+## 2018-01-15
+
+## __*Agenda*__
+* GSoC
+
+### *__Minutes__*
+* GSoC
+   * OpenCV has officially applied to be an Org for 2018!
+      * The [idea page is here](https://github.com/opencv/opencv/wiki/GSoC_2018)
+   * Optimizing nets ... can be hard
+      * Mobile net, u net for segmentation, for example VGG replaced by Mobile
+      * Replace layer by layer with quantization, pruning
+         * Fine tuning ... need to train
+      * Hardnet descriptors
+      * Features
+   * Add support for ONNX ... 
+      * Translating ONNX structures to be run in DNN
+      * TensorFlow uses different memory layout than caffe
+          * Smaller semantic differences
+      * We have a good mentor for this, if we can find the student
+      * MXnet ?
+          * Generic compilation inside of graph
+          * Halide part to generate ops and scheduling 
+          * Python interface Keras API, CNTK
+* Possible source of imagery from ...
+   * Judge appropriateness or not of images
+   * Judge affect of images on users
+   * General applicability
+      
+
+### *__To Dos__*
+* Gary
+  - [ ] Org setup 
+  - [ ] Pay
+  - [ ] Ask Yann about students for ONNX task
+
+# Notes
+
+
 <pre>
 <><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
 </pre>