I'm new to image proceesing and object recognition..From a hand drawn image, how to determine the level of correctness compared to the below image using image processing. Is there any way to do this using image processing or ML? 
Asked
Active
Viewed 161 times
0
santhosh136
- 145
- 1
- 1
- 8
-
1Do you have a dataset? Can these samples be artificially generated? – Mateen Ulhaq Jul 12 '20 at 07:45
-
You could use some combination of circle and line recognition, as well as handwritten digit recognition (e.g. MNIST). – Mateen Ulhaq Jul 12 '20 at 07:46
-
Yes I have around 70 normal hand drawn images like this. Do I need to train it? If so please tell me how to do it? – santhosh136 Jul 12 '20 at 09:10
-
70 is not a very big data set... but I guess you can apply "data augmentation" to a reasonable degree here (e.g. skewing, rotation, brightness/contrast, ...). I would start by picking a reasonably simple image classification model (e.g. ResNet-34). Perhaps also consider starting with pre-trained weights to "transfer learn" with. – Mateen Ulhaq Jul 12 '20 at 20:49