I often encounter the following graph in tensorboard, where there is a significant drop in the first couple of iterations and much slower convergence later on.
Is there a way to adjust the vertical axis range so that I can focus on the later part to see whether it is decreasing? Alternatively, can I tell tensorboard to ignore the first few data points and auto-adjust the vertical axis for the others?
As far as I know, aside from creating a whole new variable in your TF script, currently your only option is to click & drag a box around the region of interest. I found this to be slow on Firefox, but smooth on Chrome.
Related
I'm using matplotlib in spyder. Size of plot is not good enough. I can only see it in the console on the bottom right. Is there any way to make it large?
Spyder images are supposed to be adjustable based on your setup. You can do this manually through the gui. I am guessing that you are not setting your figure size which is one way to increase the size of your figure.
fig = plt.figure(figsize=(16,16)) #Larger the numbers, the larger the figure
Another thing to consider when saving your figure. Use a vectorized format like a pdf and set your dpi so that you can zoom without degrading quality. This will make the file larger of course but it helps make publication quality figures.
I want to make the plot smaller, so it can save some space in print. However, when I make it smaller, the grid will be too tight:
Honestly, I don't get it. Why can't matplotlib auto-detect that this image is too small that the lable should be with a bigger interval?
So, How can I make the grid more sparse, so the lables won't overlap with each other?
Additionally, the title will also collapse with the labels. I can set it manually, by plt.suptitle(title, y=1.08). So it will look better
But I'm wondering if there is anyway to do it automatically?
Note: I'm using seaborn, but I tried with pure matplotlib, same problem.
I've been working on implementing a real time Core Plot graph into my application on OS X. To my dismay I noticed a fairly significant issue. Once the line gets to the end of the X-Axis and it starts scrolling to keep up with the line the CPU load hits 30-35% non-stop.
I figured before I proceed any further I had better go back and see if I had made some type of mistake in my code for the CPU to spike like that. There wasn't anything out of the ordinary that I noticed, and I tried to adjust the framerate and updating frequency but without luck. I decided to go back to the real time example project they include and it has the same effect on the CPU.
Is there anything I can do about this, or is that just the nature of
real time graphing on OS X?
. .
Everything is fine for the first 50 frames (indicated by the line with arrows), but once it gets to the end of it that's where things turn for the worse.
Side Note:
I noticed Swift does graphing in the Playground, and even though it's apparently not real time (and I'm using Obj-C) it looks really sharp. Is the Swift graphing feature only available within playgrounds, or is there a way to implement that into a project? I'm only mentioning this because I'm looking to find something soon that is efficient.
That's the expected behavior with Core Plot. Once the graph starts to scroll, it has to redraw the plot, both axes, and all of the grid lines for each animation frame. You could reduce the drawing load by decreasing the number of grid lines and/or axis tick marks.
The playground graphs are a private part of the playground environment.
With imagesegmentation I segmented objects from a webcam. But because of the bad light I get a lot of noise in the picture. I want now to improve the shape of the found objects. The only method i found is image opening and closing but the result is no as good as I wished. Does anyone know some other methods?
In this folder I have the orginal image after the segmentation, the image after closing, and a picture of what kind of picture I'm looing for.
Thanks in advance
You could try the opening-closing combination operator. This provides a better smoothing of the shapes. Further more varying the radii of opening and closing (called the ASF - alternating sequential filtering) produces smoother results.
There are a few demos online that use fourier desciptors. Its also important to choose a good shape of structuring elements, since your closing seems to close convex holes in the image.
If it is salt/pepper noise, a simple median filter may be better.
I have a Core-Plot Graph within a Mac Application. But the inside table is appearing shifted down and to the left of the containing "frame/border" so that neither of the axis' are showing. I cannot figure out how to change this does anyone know what parameters I need to change to fix this?
Update: Sorry I did not realize I could upload an image, I have done so know and will try your suggestions in the mean-time.
Not sure what you mean by "inside table". It would be easier to offer suggestions if you could post a screenshot.
Without seeing what's wrong, here are some common areas to look at:
If you haven't already done so, look at the example apps included with Core Plot for ideas. The Plot Gallery app has many sample plots and the others are useful, too.
You may need to add padding (paddingLeft, paddingBottom, etc.) on the graph and/or plot area frame. Padding the graph pushes everything in away from the edges of the graph. Padding the plot area frame pushes the plot area in so the axes and titles can hang outside the plot area (the area where the plots are drawn).
If you want to keep an axes pinned to a specific place, e.g., the edge of the graph, set up a floating axis. Otherwise make sure the orthogonal coordinate is set (it defaults to 0). For example, the orthogonal coordinate for the x-axis is the y-value where the x-axis crosses the y-axis.