mean that, I have a set of images taken by Canon DSLR and other on mobile camera 12MP . I want to train a LeNet model. Can i use both set as a single dataset
We can only train a CNN model on the same size images. So, you have to resizing images for handle it
Related
I am working with the 2015 BRATS Dataset of MRI Brain Tumors. The files in the dataset are in .mha format and have many slices of 2d images that make up a 3d brain. Is there a way I can train the model on these images or do I have to convert them somehow? If so, how do I convert them?
A normal RGB image is already a 3 dimensional input. So you can just stack all the images and instead of maybe 3 channels you have 30 (for 10 stacked rgb images). You can process this data like you would process any other imgae.
What you can also try with this kind of data is using 3D convolutions on them. This way you have a 3 dimensional kernel like 3x3x3 (maybe even more in channel direction) and you slide it over the input in x,y and channel direction. This can help and boost the performance but will increase the runtime.
I'm using the wpf map control of DevExpress 15.1.5 and I'm trying to load a map from a local source.
DevExpress's help allows me to download a .osm file of a selected map area by me, but I can not load a .osm with this control; it only supports .kml and .shp.
What can I do?
You need to download map data in a different format, see. :
OpenStreetMap data
I recommend the first download a small country and try it if that's what you need.
Map consists of multiple layers, which together form "what you see". Layers are buildings, roads, places, waterways,landuse ... and are formed by shapes *.shp file (eg. Building plan, the route of the road ...)
You can view video here:
https://www.youtube.com/watch?v=4vlEzX9Q15Y&t=6s
also the source code is available in github.
the Control works on local OSM tiles
it is written in C#
I created OSM MAP Control which have the following functionalities:
1. Accurate Zoom with the mouse wheel.
2. Move left , up ,bottom and right
3.if tiles are missing , it ask to start download those missing tiles.
4. Create line on the map
5.save history by friendly name
6. you can download area of tiles - with zoom.
We have a system in which we have big plan images loading in multiscale deepzoom images like an image gallery.Images are usually of very high resolutions.
Users have the ability to draw shapes ect on selected deepzoom image.We are using a canves on deepzoom image and save all overlay shapes drwan along with their logical poistion in db.thus the drawn shapes auto sets their position as user pans in or pans out.
We need an automated solution so we can take the printout of all the plan images along with all overlay work done on each deepzoom image.
Currently we are doing this one by one i.e we use Grid,Multistage,Canvas.We load image in Multiscale image,then draw Canves then we use following code to same them in jpeg from Grid(ContentGrid)
SaveFileDialog saveDlg = new SaveFileDialog();
saveDlg.Filter = "JPG Files (*.jpg|*.jpg|*.jpeg|*.jpeg)";
saveDlg.DefaultExt = ".jpg";
if ((bool)saveDlg.ShowDialog())
{
using (Stream fs = saveDlg.OpenFile())
{
SaveToFile(new WriteableBitmap(ContentGrid, null), fs);
}
}
But this solution does not work properly.This throws memory issues when deepzoom image is big one.
We want something like this.For all plans images in gallery we need to loop and programatically convert to jpeg along with canves etc and then save to Amazon s3 by program.An automation will pick all the jpegs and make a single pdf for all those.
or an automated solution.
Please suggest.
I have an image in Wpf and I want to print it in scale 1:1.
The printer can be any printer (Well it has to be able to print graphics).
How to I do this?
If the image is larger than the printers page size, is it possible to split the image into several pages? How?
// Anders
i want to implement 3D image manipulation(i.e., image edit,re-size,rotate,zoom-in & zoom-out).
please give me your ideas as soon as possible.
Silverlight 2.0 does not support true 3D. Silverlight 3 will support a perspective projection simulated 3D (think 3D images in PowerPoint).
For standard image manipulation you have a few ways to go. If you just want to do rotation and resizing for display purposes, you can set a RenderTransform on the image object. This does not actually change the bits in the image but just cases them to be drawn differently.
If you want to learn about full image processing, see this other post on StackOverflow:
Image Processing