Daniel J. H.

401 Unauthorized for Emerald Base Map

1

It seems like the API disallows some default map ids. Is this a server limitation, or simply how wallmapper accesses the map tiles? Example: ``` daniel@x1c /t/wallmapper> python wallmapper.py --zoom...

- [ ] Bump packages for both cpu as well as gpu docker images. - [ ] Re-calculate dependencies and lock dependency tree via pip-compile. - [ ] Test end...

Generalizes post-processing handlers across zoom levels, closes #51

1

For #51. WIP, and needs testing. See https://wiki.openstreetmap.org/wiki/Slippy_map_tilenames#Resolution_and_Scale

Implements topology-aware loss function; resolves #133

6

First try at a topology-aware loss function - https://arxiv.org/abs/1712.02190 - https://arxiv.org/abs/1603.08155 The idea is to optimize for topology like connectivity directly instead of just for pixel-wise loss. Here is an...

Adds squeeze and excitation (scSE) modules, resolves #157

7

For https://github.com/mapbox/robosat/issues/157. Adds scSE modules :boom: :rocket: https://arxiv.org/abs/1709.01507 > Squeeze-and-Excitation Networks https://arxiv.org/abs/1803.02579 > Concurrent Spatial and Channel 'Squeeze & Excitation' in Fully Convolutional Networks  _from https://arxiv.org/abs/1803.02579_ Tasks - [x]...

Adds feature pyramid attention (FPA) module, resolves #167

1

For #167. Adds Feature Pyramid Attention (FPA) module :boom: :rocket: > Pyramid Attention Network for Semantic Segmentation https://arxiv.org/abs/1805.10180  _from https://arxiv.org/abs/1805.10180 Figure 2_  _from https://arxiv.org/abs/1805.10180 Figure 3_ Tasks -...

Benchmark using pinned CUDA memory in data loaders

4

At the moment the data loaders load up images from the dataset, do pre-processing (like normalization), and then convert the images into tensors. Then we copy the data from CPU...

Implements Feature Pyramid Network

5

For https://github.com/mapbox/robosat/issues/60. This changeset implements a Feature Pyramid Network (FPN) on top of a (potentially pre-trained) ResNet. - [Feature Pyramid Networks for Object Detection](https://arxiv.org/abs/1612.03144) - [A Unified Architecture for Instance...

Uses EfficientNetB0 as segmentation model encoder backbone

1

For https://github.com/mapbox/robosat/issues/172 (see for context) - this implements the EfficientNetB0 model as an encoder for our encoder-decoder architecture. I'm currently training my EfficientNet model family (no h-swish, no squeeze-and-excitation) in...

Implement EfficientNet backbone

2

This recently came out https://arxiv.org/abs/1905.11946 > EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks with incredible practical implications: we can either get a much better backbone than our current resnet50...