 |
.. (לתיקייה המכילה) | |
In Question 1, Section B, how do I calculate the recall when the edge can be 3 pixels far in the Groundtruth image? | |
|
Remember that you must make the Groundtruth image binary in some reasonable way for your calculations. The intersection in the numerator should be with a modified Groundtruth image that should reflect the 3 pixel detection. The denominator should be the original Groundtruth image. |
In Question 2, Section A, how do I normalize certain color spaces to [0,1]? | |
| Some color spaces can't be normalized easily. It is ok not to normalize if you don't know how to do it. |
In Question 2, Section B-6, what do you expect as an answer? | |
|
The question here is actually about the colors which the camera is sure to represent with a certain error. Think of the geometric location of all the colors in the CIE diagram that are sure to be affected due to the error in the green sensor, no matter what the actual output capabilities of the camera. In this question we are not asking about the practical output of the camera, but what colors will definitely have a representation error disregarding the camera actual output. The answer should be a geometric location on the diagram of all such colors, or it's complement. |
Can I use the Matlab backslash operator in Question 4? | |
|
The backslash operator practically solves the Least Squares problem. You may use it to verify that your solution is valid and correct, but you may NOT use it to skip calculations. We require that you show all calculation steps in reaching the final answer. |
How can I solve the non-linear model in Question 4, Section B? | |
| You can solve it by making it linear, and then using the tools we learned in the tutorial. |