CNN-Ecognition
Dear sir/ madam, I am new to the object based image analysis (OBIA). In this connection, I initially tried to extract road and building using high resolution data using ecognition. But i am facing difficulties while extracting using Deep learning (convolution neural network) techniques. Misclassification was noticed between this two features (Building and road), because similar tonal signature and shadows from building and trees. Subsequently I tried to insert the shape of this features to discriminate (Because modifications are noticed in shape of this both features)in the sample file. But all the generated samples are only in defined shape (kernal in size (for example 10 X 10 pixels). Hence, I unable to define the shape of the feature too. I also tried to generate sample files manually in different shapes and replaced the sample files, but its not consider while processing. So kindly guide me to do further for extracting road and building using deep learning in Ecognition.
- Dr.
- 01 Jan
- 0 Answer
0 Answer
0 Answer
Other Related Discussions
Didn't get the answer.
Contact people of Talent-Neural Networks directly by clicking here
