Trainable Weka Segmentation
- We continued to familiarize ourselves with the Weka plugin by using different techniques and different filters.
- At first, I hand traced each tissue segment hoping that the plugin would be able to identify each part in steem
-
Issue: The plugin was not able to segment each tissue and it ended up not working all together. My computer froze and crashed which led me to believe that Weka failed when I hand traced it because it had to read large amounts of data from the outlined areas. -
Success: I had success being able to isolate the Fiber tissues, the cortex and phloem tissues (for the most part). Our goal is to be able to isolate the pith, xylem, phloem, fiber, and cortex.Issues: I was not able to isolate the xylem itself as it was identified by Weka as being similar to the pith (they were all the same color).
-
Success: The segmentation of this trial was much more successful than the previous two trials. I was able to differentiate the background, the cortex, fiber tissues, phloem, xylem and pith with different colors.
Issues: I could perfect the segmentation a little bit more since the plugin sees small areas of the pith as part of the background.
Nikon Specialist
- The people who installed the new Nikon Cameras (Nikon TE-DH100W paired with the Nikon DS-U3 & DS-Fi1) in Dr. Turners lab taught us how to fix shadowing issues from our previous images using the NIS-Element Software.
- By fixing the shadowing issues as we acquired the images on the scope, our images would look clearer and the lighting of the stitched images would be consistent. This would eliminate variables that could alter our area measurements within the stem. With NIS-Element software, we are also able to a completely stitched image without using imagej.