Our projects:


Unofficial implementation of Unsupervised Monocular Depth Estimation neural network MonoDepth in PyTorch. The network is designed to produce high quality depth predictions from a single image after learning from a training dataset of stereo images.


Kaggle Ship Detection Challenge Solution

This project was an entry for the Airbus Ship Detection Challenge held on Kaggle. It implemets all steps of the solution: from assembling datasets to training the neural network and submitting the final file. The developed functionality can be used as a universal high-quality baseline solution for any segmentation task.


Semantic Segmentation with MobileNet v3

Customizable, module based implementation of MobileNetV3 in Tensorflow (and TensorflowLite). This architecture may be used for semantic segmentation as well as for classification.


Semantic Segmentation of Seismic Reflection

Deep learning based semantic segmentation of reflection seismology images for salt deposits recognition.


© 2019-2024 OniroAI (former ClubAI)