StreetVision Roadwork

Binary image classifier prepared for NATIX StreetVision subnet inference.

Labels

  • 0: None
  • 1: Roadwork

Current status

This repository currently contains a baseline copied from reyraa/sn72-roadwork. It has not yet been fine-tuned by FavoritCoin and no accuracy claim is made until evaluation on a labeled roadwork/no-roadwork benchmark is complete.

Verified runtime

Tested with StreetVision commit 5d0756bfb5c827929b02d66446b261fcb14c063a on NVIDIA RTX 4070 Ti.

  • Output is a probability in the 0.0-1.0 range.
  • Model loads through AutoModelForImageClassification.
  • GPU inference is approximately 4.13 ms/image after warmup.
  • GPU memory allocation is approximately 336 MiB.

Files

  • config.json
  • preprocessor_config.json
  • model.safetensors

Limitations

Training data provenance and benchmark accuracy need independent validation before production mining or registration.

Downloads last month
74
Safetensors
Model size
85.8M params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support