- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
For those interested in object detection with the NCS, there is now an example of Tiny YOLO with NCS in the ncappzoo here: https://github.com/movidius/ncappzoo/tree/master/caffe/TinyYolo
After cloning the repository you'll find the example in the ncappzoo/caffe/TinyYolo directory. Inferences with this network will give you object classifications along with rectangles that specify where the objects are within the original image. See the project's README.md file for more information.
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Awesome. Thanks for this well-commented and annotated sample; cleared up a lot for me.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
is there an example for Tensorflow implementation?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
is there an example for Tensorflow implementation? Yes there is.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
May you direct me because i cant find it in the repo.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
When you install the Movidius tools from Git Hub into workspace there are sub folders.
One of the folders is called InceptionV3.InceptionV2 Those are Tensorflow examples.
On the flip side, as I know there is no information on the end to end process. Which is
Create a graph
build the graph to export files ( from what I can tell you need two files) look at the arguments for
mvcscompile
load those files and compile the files
Now run that result on your stick.
I dont see that kind of end to end thing anyplace and am trying to figure it out as well.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
By the way YOLO is far more interesting since it delivers several inferences simultaneously and bounding regions.
I think this is what most people think of when they think of object recognition unless they are working on identifying
things on an assembly line, one at a time.
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page