• Stars
    star
    208
  • Rank 187,913 (Top 4 %)
  • Language
    Python
  • Created over 7 years ago
  • Updated over 6 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Kaggle DSTL Satellite Imagery Feature Detection

DSTL

https://www.kaggle.com/c/dstl-satellite-imagery-feature-detection/

Note: this repo is not supported. License is MIT.

Object types

Note that labels here are 1 less than in submission file:

  • 0: Buildings - large building, residential, non-residential, fuel storage facility, fortified building
  • 1: Misc. Manmade structures
  • 2: Road
  • 3: Track - poor/dirt/cart track, footpath/trail
  • 4: Trees - woodland, hedgerows, groups of trees, standalone trees
  • 5: Crops - contour ploughing/cropland, grain (wheat) crops, row (potatoes, turnips) crops
  • 6: Waterway
  • 7: Standing water
  • 8: Vehicle Large - large vehicle (e.g. lorry, truck, bus), logistics vehicle
  • 9: Vehicle Small - small vehicle (car, van), motorbike

General approach

UNet network with batch-normalization added, training with Adam optimizer with a loss that is a sum of 0.1 cross-entropy and 0.9 dice loss. Input for UNet was a 116 by 116 pixel patch, output was 64 by 64 pixels, so there were 16 additional pixels on each side that just provided context for the prediction. Batch size was 128, learning rate was set to 0.0001 (but loss was multiplied by the batch size). Learning rate was divided by 5 on the 25-th epoch and then again by 5 on the 50-th epoch, most models were trained for 70-100 epochs. Patches that formed a batch were selected completely randomly across all images. During one epoch, network saw patches that covered about one half of the whole training set area. Best results for individual classes were achieved when training on related classes, for example buildings and structures, roads and tracks, two kinds of vehicles.

Augmentations included small rotations for some classes (±10-25 degrees for houses, structures and both vehicle classes), full rotations and vertical/horizontal flips for other classes. Small amount of dropout (0.1) was used in some cases. Alignment between channels was fixed with the help of cv2.findTransformECC, and lower-resolution layers were upscaled to match RGB size. In most cases, 12 channels were used (RGB, P, M), while in some cases just RGB and P or all 20 channels made results slightly better.

Validation

Validation was very hard, especially for both water and both vehicle classes. In most cases, validation was performed on 5 images (6140_3_1, 6110_1_2, 6160_2_1, 6170_0_4, 6100_2_2), while other 20 were used for training. Re-training the model with the same parameters on all 25 images improved LB score.

Some details

  • This setup provides good results for small-scale classes (houses, structures, small vehicles), reasonable results for most other classes and overfits quite badly on waterway.
  • Man-made structures performed significantly better if training polygons were made bigger by 0.5 pixel before producing training masks.
  • For some classes (e.g. vehicles), it helped a bit to make the first downscaling in UNet 4x instead of default 2x, and also made training 1.5x faster.
  • Averaging of predictions (of one model) with small shifts (1/3 of the 64 pixel step) were used for some classes.
  • Predictions on the edges of the input image (closer than 16 pixels to the border) were bad for some classes and were left empty in this case.
  • All models were implemented in pytorch, training for 70 epochs took about 5 hours, submission generation took about 30 minutes without averaging, or about 5 hours with averaging.

Other things tried

A lot of things that either did not bring noticeable improvements, or made things worse:

  • Losses: jaccard instead of dice, trying to predict distance to the border of the objects.
  • Color augmentations.
  • Oversampling of rare classes.
  • Passing lower-resolution channels directly to lower-resolution layers in UNet.
  • Varying UNet filter sizes, activations, number of layers and upscale/downscale steps, using deconvolutions instead of upsampling.
  • Learning rate decay.
  • Models: VGG-like modules for UNet, SegNet, DenseNet

Object types stats

Area by classs:

im_id 0 1 2 3 4 5 6 7 8 9
6010_1_2 0.0% 0.0653% 0.0% 1.3345% 4.5634% 0.0% 0.0% 0.0% 0.0% 0.0%
6010_4_2 0.0% 0.0% 0.0% 1.9498% 12.3410% 0.0% 0.0% 0.0% 0.0% 0.0%
6010_4_4 0.0% 0.0% 0.0% 0.0% 22.8556% 0.0% 0.0% 0.0% 0.0% 0.0%
6040_1_0 0.0% 0.0% 0.0% 1.4446% 8.0062% 0.0% 0.0% 0.0% 0.0% 0.0%
6040_1_3 0.0% 0.0% 0.0% 0.2019% 18.7376% 3.6610% 0.0% 0.0% 0.0% 0.0%
6040_2_2 0.0% 0.0% 0.0% 0.9581% 18.7348% 0.0% 0.0% 0.0% 0.0% 0.0%
6040_4_4 0.0% 0.0% 0.0% 1.8893% 2.9152% 0.0% 0.0% 0.0% 0.0% 0.0%
6060_2_3 0.1389% 0.3037% 0.0% 3.0302% 8.4519% 93.5617% 0.0% 0.0% 0.0% 0.0003%
6070_2_3 1.5524% 0.3077% 0.8135% 0.0% 16.0439% 0.0% 10.6325% 0.0543% 0.0% 0.0058%
6090_2_0 0.0% 0.0343% 0.0% 0.4072% 10.1105% 28.2399% 0.0% 0.3130% 0.0% 0.0008%
6100_1_3 8.7666% 2.7289% 2.2145% 12.2506% 6.2015% 2.6901% 0.0% 0.6839% 0.0110% 0.0459%
6100_2_2 3.1801% 0.8188% 1.1903% 3.7222% 7.6089% 44.3148% 1.8823% 0.0512% 0.0100% 0.0242%
6100_2_3 8.2184% 1.4110% 1.2099% 9.5948% 7.5323% 0.0% 0.0% 0.0603% 0.0148% 0.0661%
6110_1_2 13.1314% 2.8616% 0.4192% 4.1817% 3.3154% 49.7792% 0.0% 0.1527% 0.0% 0.0065%
6110_3_1 4.5495% 1.2561% 3.6302% 2.8221% 5.4133% 57.6089% 0.0% 0.5531% 0.0181% 0.0253%
6110_4_0 2.4051% 0.5732% 1.8409% 2.8067% 5.7379% 80.7666% 0.0% 1.4210% 0.0136% 0.0017%
6120_2_0 1.7980% 0.7257% 0.8505% 4.4026% 5.6352% 79.5910% 0.0% 0.0% 0.0138% 0.0041%
6120_2_2 20.6570% 2.0389% 4.2547% 8.6533% 4.4347% 10.2929% 0.0% 0.2859% 0.0076% 0.1560%
6140_1_2 12.9211% 2.4488% 0.3538% 4.1461% 3.1027% 49.5910% 0.0% 0.1415% 0.0% 0.0086%
6140_3_1 5.2015% 1.4349% 3.4252% 2.5189% 5.8852% 57.3959% 0.0% 0.4664% 0.0042% 0.0358%
6150_2_3 0.0% 0.6055% 0.0% 3.0197% 13.5187% 80.6649% 0.0% 0.0% 0.0% 0.0%
6160_2_1 0.0% 0.0% 0.0% 2.7986% 10.2713% 0.0% 0.0% 0.0% 0.0% 0.0%
6170_0_4 0.0% 0.0016% 0.0% 0.1994% 24.8913% 0.0% 0.0% 0.0152% 0.0% 0.0%
6170_2_4 0.0% 0.0011% 0.0% 2.5070% 7.7844% 49.5326% 0.0% 0.0089% 0.0% 0.0%
6170_4_1 0.0% 0.0% 0.0% 0.1349% 20.2214% 0.0% 0.0% 0.0% 0.0% 0.0%

Making a submission

Train a CNN (choose number of epochs and other hyper-params running without --all):

$ ./train.py checkpoint-folder --all --hps dice_loss=10,n_epochs=70

Make submission file (check hyperparameters doing a submission for the model trained with validation by running with --validation *value* and optionally --valid-polygons):

$ ./make_submission.py checkpoint-folder submission.csv.gz

Finally, use ./merge_submission.py to produce the final submission.

This just gives a general idea, real submissions were generated with different hyperparameters for different classes, and all above commands have more options that are documented in the commands themselves (use --help, check the code if in doubt).

More Repositories

1

transformer-lm

Transformer language model (GPT-2) with sentencepiece tokenizer
Python
161
star
2

kaggle-dsbowl-2018-dataset-fixes

Kaggle Data Science Bowl 2018 dataset fixes
Python
131
star
3

kaggle-script-template

Kaggle script build system template
Python
112
star
4

kaggle-imet-2019

Python
91
star
5

kaggle-kuzushiji-2019

Kaggle Kuzushiji Recognition: 2nd place solution
Python
80
star
6

python-adagram

AdaGram (adaptive skip-gram) for Python
Python
75
star
7

scrapy-pyppeteer

Use pyppeteer from a Scrapy spider
Python
60
star
8

kaggle-lions-2017

2nd place solution for Kaggle NOAA Fisheries Steller Sea Lion Population Count
Jupyter Notebook
36
star
9

tf-rnn-char-autoencoder

RNN Character Autoencoder built with TensorFlow
Python
23
star
10

mapillary-vistas-2017

WIP
Python
20
star
11

kaggle-rsna-2019

https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/
Python
18
star
12

kaggle-jigsaw-2019

Python
14
star
13

kaggle-panda-2020

Python
7
star
14

kings-problem

Calculate the number of non-attacking kings placements of n*n kings on 2*n x 2*n board
C++
5
star
15

kaggle-fashion-2019-mm

Python
4
star
16

gram-matcher

Match lexico-grammer templates with natural language sentences, using pymorphy
Python
4
star
17

tpu-imagenet

Train on ImageNet on Kaggle TPUs
Python
4
star
18

sensefreq

Sense frequencies and WSD
Python
4
star
19

kaggle-amazon-2017

Solution for Kaggle challange "Planet: Understanding the Amazon from Space"
Jupyter Notebook
4
star
20

rust-broad-crawl

A very basic broad crawler in rust
Rust
3
star
21

ru-jungle

Python
3
star
22

python-emacs-stuff

Various python emacs stuff
Emacs Lisp
2
star
23

ruslang-wsd-labeled

Labeled contexts of Russian polysemous words
Python
2
star
24

vmprofit

VMprof helpers
Python
2
star
25

symbmath

Symolic mathematics in C++ - diffirentiating, expression simplification, integrating
C++
2
star
26

3dbilliard

JavaScript
1
star
27

interactive-opengl

Interactive OpenGL console (educational)
Python
1
star
28

trac-ajaxcomments

Make trac ticket page more convinient: post, update and merge comments without reloading the page
JavaScript
1
star
29

SmartSketcher

A drawing tool that aims at replacing a notepad for doing creative work
Java
1
star
30

wp-doc-gen

Python
1
star
31

clj-grantt

Planning work described by grantt diagram, according to given constraints
Clojure
1
star
32

web-adagram

AdaGram model on the web
Python
1
star
33

meteor-bugtracker

Simple BugTracker app - testing Meteor framework
JavaScript
1
star
34

dbfpy

porting to python3
Python
1
star
35

WSI

Word Sense Induction
Jupyter Notebook
1
star
36

imagenet-exp

ImageNet experiments
Python
1
star
37

sense-grouping-survey

Sense grouping survey
Python
1
star
38

dynide

Python Dyncamic IDE - test and re-run your code without leaving your favorite editor
Python
1
star