This produce application detects undesired materials in tomatoes. The material-other-than tomatoes (m.o.t.) is trained on (with LEARN), and then the tomatoes are forgotten (with FORGET). The tomato stems, which are considered OK, are forgotten also. In this way, other acceptable materials may be easily trained as OK. After several examples of m.o.t. are trained, the system is ready to detect these contaminants.
We use the new and powerful Coloration learning mode to demonstrate how easy, straightforward, and effective our trainable vision is.
We start by training on some example contaminants - the small orange-colored dots show learning activity:
|
 |
 |
 |
 |
 |
 |
 |
Forgetting (FORGET) tomatoes - the small blue dots represent forgetting activity.: |
Notice tomato stem is detected as contaminant - the bright green color shows detection: |
 |
 |
To make tomato stem OK, we forget (FORGET) it - makes both tomatoes and stem are OK: |
System now treats tomato stem as OK, but still detects contaminants: |
 |
 |
Note that a dark leaf, a contaminant, is not detected: |
To add it to the repertoire of contaminants, we train on a leaf example with LEARN: |
 |
 |
Now, after training, the leaf contaminant is easily detected: |
The system still detects sticks, rocks, but not tomato stems. We can train one contaminant after another into the PC-Eyebot's detection ability. We can also forget various kinds of tomatoes, making it pass as OK a wide variety - all while maintaining the ability to detect the desired contaminants. |
 |
 |