We are looking for a tool or script that is able to read in images from a source directory and to:
1. Separate and move the images from the original source directory to a new set of directories, whose name is based on whether nudity or amounts of skin tone are present in the image or not: e.g. No nudity 0-20% skin tone, Some nudity 21-60% skin tone, or Lots of Nudity 61%+
2. Produce a simple log file identifying the image file name, and which directory / category it has been placed into. This log must also include the percentage amount of skin tone and nudity found in the image
- Application preferably written in Java, but open to interpretation, as long as it can be integrated into larger application
- Preferably with an API
- Code and functions must be commented and plain text (i.e. not obfuscated)
- Basic guide on how to run the code must be included
- Work on Windows as a minimum, but ideally Mac OS too
- The developer must provide a working example of the program including legal test images
- Please note: this must be a local / desktop tool - not an online or website tool
Must explain which nudity recognition technique has been used and how it works (or at least provide the literature for further review)
NOTE: We are not looking for a GUI development, although if it makes it easier to run a script via a very basic GUI then that’s OK with us.
The use of 3rd party code is acceptable but must be indicated to us and must allow us to use/negotiate for commercial usage in the future.