CALTECH101

CALTECH101

The CALTECH101 dataset

Caltech-101 contains a total of 9,146 images, split between 101 distinct object categories (faces, watches, ants, pianos, etc.) and a background category. This dataset contains 102 folders, the BACKGROUND_Google (the background category) can be removed, and users may use the left 101 categoies.

Dataset Statistics

  1. Color: RGB
  2. Sample size: Roughtly 300x200
  3. Dataset size: 1.2 GB

Overall, the dataset consists of pictures of objects belonging to 101 categories. About 40 to 800 images per category. Most categories have about 50 images. The size of each image is roughly 300x200 pixels. Almost all images are annotated with the following information: a bounding box of the object, and a carefully traced silhouette of the objects by a human subject.

The Number of Samples per Category for Caltech 101

TODO

Samples

CALTECH101 Sample

Dataset Usage

Theano

ini_caltech101:

Getting the Code

To get a local copy of the code, clone it using git:

git clone https://github.com/marcuniq/ini_caltech101.git
cd ini_caltech101

Make sure you have the bleeding edge version of Theano, or run

pip install --upgrade --no-deps git+git://github.com/Theano/Theano.git

Next, install the package. Use ‘develop’ instead of ‘install’ if you consider changing package code

python setup.py develop

Run train.sh (sets proper theano env flags), which downloads and untars the ‘img-gen-resized’ dataset, then starts training.

./train.sh

Keras

deeplearn-caltech101: Build on top of VGG19

The dataset is already contained in this project.

pip install -r requirements.txt

python2.7 caltech_convnet.py

Deeplearning4j

Caltech101Classifier: VGG16 pretrained model.

How to run the project:
java -jar -Xms8g -Xmx8g Caltech101Classifier-1.0-SNAPSHOT-jar-with-dependencies.jar <data path>

Reference:

  1. Images only: L. Fei-Fei, R. Fergus and P. Perona. Learning generative visual models from few training examples: an incremental Bayesian approach tested on 101 object categories. IEEE. CVPR 2004, Workshop on Generative-Model Based Vision. 2004

  2. Images and annotations: L. Fei-Fei, R. Fergus and P. Perona. One-Shot learning of object categories. IEEE Trans. Pattern Recognition and Machine Intelligence. In press.

Specific Categories:

  1. trilobite
  2. face
  3. pagoda
  4. tick
  5. inlineskate
  6. metronome
  7. accordion
  8. yinyang
  9. soccerball
  10. spotted cat
  11. nautilus
  12. grand-piano
  13. crayfish
  14. headphone
  15. hawksbill
  16. ferry
  17. cougar-face
  18. bass
  19. ketch
  20. lobster
  21. pyramid
  22. rooster
  23. laptop
  24. waterlilly
  25. wrench
  26. strawberry
  27. starfish
  28. ceilingfan
  29. seahorse
  30. stapler
  31. stop-sign
  32. zebra
  33. brontosaurus
  34. emu
  35. snoopy
  36. okapi
  37. schooner
  38. binocular
  39. motorbike
  40. hedgehog
  41. garfield
  42. airplane
  43. umbrella
  44. panda
  45. crocodile-head
  46. llama
  47. windsor-chair
  48. car-side
  49. pizza
  50. minaret
  51. dollarbill
  52. gerenuk
  53. sunflower
  54. rhino
  55. cougar-body
  56. crab
  57. ibis
  58. helicopter
  59. dalmatian
  60. scorpion
  61. revolver
  62. beaver
  63. saxophone
  64. kangaroo
  65. euphonium
  66. flamingo
  67. flamingo-head
  68. elephant
  69. cellphone
  70. gramophone
  71. bonsai
  72. lotus
  73. cannon
  74. wheel-chair
  75. dolphin
  76. stegosaurus
  77. brain
  78. menorah
  79. chandelier
  80. camera
  81. ant
  82. scissors
  83. butterfly
  84. wildcat
  85. crocodile
  86. barrel
  87. joshua-tree
  88. pigeon
  89. watch
  90. dragonfly
  91. mayfly
  92. cup
  93. ewer
  94. octopus
  95. platypus
  96. buddha
  97. chair
  98. anchor
  99. mandolin
  100. electric-g