visionscript 0.0.2

VisionScript is an abstract programming language for doing common computer vision tasks, fast.

VisionScript

VisionScript is an abstract programming language for doing common computer vision tasks, fast.

VisionScript is built in Python, offering a simple syntax for running object detection, classification, and segmentation models. Read the documentation.

Get Started

First, install VisionScript:

pip install visionscript

You can then run VisionScript using:

vscript

This will open a VisionScript REPL in which you can type commands.

Run a File

To run a VisionScript file, use:

vscript ./your_file.vic

Use VisionScript in a Notebook

VisionScript offers an interactive web notebook through which you can run VisionScript code.

To use the notebook, run:

vscript notebook

This will open a notebook in your browser. Notebooks are ephermal. You will need to copy your code to a file to save it.

Quickstart

Find people in an image using object detection

Load["./photo.jpg"]
Detect["person"]
Say[]

Find people in all images in a folder using object detection

IN["./images"]
    Detect["person"]
    Say[]

Replace people in a photo with an emoji

Load["./abbey.jpg"]
Size[]
Say[]
Detect["person"]
Replace["emoji.png"]
Save["./abbey2.jpg"]

Classify an image

Load["./photo.jpg"]
Classify["apple", "banana"]

Installation

To install VisionScript, clone this repository and run pip install -r requirements.txt.

Then, make a file ending in .vic in which to write your VisionScript code.

When you have written your code, run:

python3 lang.py --file ./your_file.vic

Run in debug mode

Running in debug mode shows the full Abstract Syntax Tree (AST) of your code.

python3 lang.py --file ./your_file.vic --debug

Debug mode is useful for debugging code while adding new features to the VisionScript language.

Documentation

• Load["./abbey.jpg"] -> Load the image
• Size[] -> Get the size of the image
• Say[] -> Say the result of the last function
• Detect["person"] -> Detect the person
• Replace["emoji.png"] -> Replace the person with black image
• Cutout[] -> Cutout the last detections
• Count[] -> Count the last detections
• CountInRegion[0, 0, 500, 500] -> Count the last detections in the region (x1, y1, x2, y2)
• Classify["cat", "dog"] -> Classify the image in the provided categories
• Save["./abbey2.jpg"] -> Save the last image
• Show[] -> Show the last image
  • If you have run inference, this will plot inference results on the image with which you are working.
• x = 1 -> Set the variable x to 1
• True and False: Booleans
• If[Statement]: If the statement is true, run the next line (the only value that evaluates to False is False right now so this is not yet useful).
• x == y: Test for equality. Check if x is equal to y.
• In["./images"]: Load all images in the ./images folder

Core Libraries Used

• CLIP
• YOLOv8
• FastSAM
• supervision
• PIL
• Lark for lexing

Inspiration

The inspiration behind this project was to build a simple way of doing one-off tasks.

Consider a scenario where you want to run zero-shot classification on a folder of images. With VisionScript, you can do this in two lines of code:

In["./images"]
    Classify["cat", "dog"]

VisionScript is not meant to be a full programming language for all vision tasks, rather an abstract way of doing common tasks.

VisionScript is ideal if you are new to concepts like "classify" and "segment" and want to explore what they do to an image.

Syntax

The syntax is inspired by both Python and the Wolfram Language. VisionScript is an interpreted language, run line-by-line like Python. Statements use the format:

Statement[argument1, argument2, ...]

This is the same format as the Wolfram Language.

Lexical Inference and Memory

An (I think!) unique feature in VisionScript compared to other languages is lexical inference.

You don't need to declare variables to store images, etc. Rather, you can let VisionScript do the work. Consider this example:

Load["./photo.jpg"]
Size[]
Say[]

Here, Size[] and Say[] do not have any arguments. Rather, they use the last input. Wolfram Alpha has a feature to get the last input using %. VisionScript uses the same concept, but with a twist.

Indeed, Size[] and Say[] don't accept any arguments.

Developer Setup

If you want to add new features or fix bugs in the VisionScript language, you will need to set up a developer environment.

To do so, clone the language repository:

git clone https://github.com/capjamesg/VisionScript

Then, install the required dependencies and VisionScript:

pip install -r requirements.txt
pip install -e .

Now, you can run VisionScript using:

python3 lang.py

Tests

Tests are run to ensure programs execute in full. Tests do not verify the output of each statement, although this will be added.

For now, you can run all test cases using the following command:

python3 test.py

Code Organization

• lang.py: Core language code.
• test.py: Run tests.
• usage.py: Variables referenced for usage instructions in lang.py.
• grammar.py: The VisionScript grammar.
• tests/: VisionScript tests.

Supported Models

VisionScript provides abstract wrappers around:

• CLIP by OpenAI (Classification)
• Ultralytics YOLOv8 (Object Detection Training, Segmentation Training)
• FastSAM by CASIA-IVA-Lab. (Segmentation)
• GroundedSAM (Object Detection, Segmentation)
• BLIP (Caption Generation)
• ViT (Classification Training)