Visual_Dataset_Generation_Basic_English_Words

---

Word Data

Source: Ogden's 850 Basic English Words

Link: http://ogden.basic-english.org/words.html

Format: lowercase text

Categories and Count:

• Operations: 100
• Things: 600
• Qualities: 150

Stored in the three text files by category names.

Visual Data Composition

• 12000 images
• 1200 images per word, 100 words

Visual Data Format

• category_word_count.png
• category abbreviation: opt, thing, qual
• count: 1-1200 per image
• 224px * 224px images
• Black word printed on white background

Data Generation

Generation Format Reference: Modeling the Visual Word Form Area Using a Deep Convolutional Neural Network - Sandy Wiraatmadja, Garrison W. Cottrell

Link: https://cogsci.mindmodeling.org/2016/papers/0435/paper0435.pdf

Data Generation Implementation Setup:

• Randomly choose 100 words from the three categories with ratio: (Operations: Things: Qualities = 12 : 70: 18)
  • Selected words will be stored in words.txt
• Randomly select 75 fonts from the font family of Python Pillow library
  • Selected fonts will be stored in fonts.txt

Data Generation Implementation:

• Random font out of the 75 fonts selected
• Random font size from [12pt, 15pt, 18pt]
• Random clockwise rotation [-15 to 15 degree]