Python Computer Vision Developer

We are looking for a Python Computer Vision Developer to build a lightweight Transformer-based system for automated product image inspection. The goal is to identify a unique problem: *detecting subtle defects that appear only under specific lighting conditions or viewing angles, even when similar defects are absent from the training dataset.* We aim to develop a simple, efficient solution using pretrained vision models and an existing dataset. *Scope of Work:* - Implement a Transformer-based model such as ViT or DINOv2 for anomaly detection. - Detect and highlight unusual regions in product images. - Evaluate the model using an existing dataset. - Develop a simple FastAPI endpoint for image uploads and predictions. - Provide source code and documentation. *Required Skills:* - Python, PyTorch, and OpenCV. - Experience with Vision Transformers or pretrained vision models. - Basic knowledge of anomaly detection. - FastAPI model integration. *To Apply:* Briefly explain which Transformer model you would use and how you would detect previously unseen defects with minimal training data. *Project Type:* One-time project with potential for future collaboration.