Seeking a skilled professional to promptly test and deploy open source models designed for diffusion image detection, such as https://arxiv.org/abs/2303.09295 and https://arxiv.org/abs/2307.06272v1 or selecting models at your discretion.
Input: Image file
Output: Binary classification indicating diffusion image generation.
Deliverables Include:
1. A deployable Docker file.
2. Inference endpoint (predict.py) that is ready for deployment on an EC2 instance. We will provide you with server.py using sanic.
Inference should complete within a 5-second limit for each image. We'll be running it on p3.2xlarge. The model's accuracy and performance should be reported on a set of 100 diffusion-based images or deepfakes benchmarks.
Please apply only if you could deliver the project by February 23rd. Given the task involves just running open-source code and containerization, we anticipate the workload to be relatively low.
In your application, please include the diffusion detection models you intend to use and an estimate of the time required.
This job is already closed and no longer accepting applicants, sorry.
