A standard U-Net with dropout and Adam is trained on 5,000 satellite landform images, yielding Dice 69.62%, but the paper provides no data, code, or held-out evaluation and overstates its comparison to prior work.
2016 IEEE 19th Interna- tional Conference on Intelligent Transportation Systems (ITSC), Rio de Janeiro, Brazil, pp
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Convolutional Neural Network Segmentation for Satellite Imagery Data to Identify Landforms Using U-Net Architecture
A standard U-Net with dropout and Adam is trained on 5,000 satellite landform images, yielding Dice 69.62%, but the paper provides no data, code, or held-out evaluation and overstates its comparison to prior work.