Deep Learning · Computer Vision

DecaVision — Image Recognition

A Wide SE-ResNet trained on CIFAR-10 that recognises ten everyday object categories from a single image.

📤   Upload your image

🎯   Prediction

Upload or paste an image to see the result.

📚   Supported categories
✈️ Airplane🚗 Automobile🐦 Bird🐱 Cat🦌 Deer🐶 Dog🐸 Frog🐴 Horse🚢 Ship🚚 Truck
  • Architecture: Wide SE-ResNet (residual blocks with Squeeze-and-Excitation attention)
  • Training data: CIFAR-10 (60,000 32×32 colour images across 10 classes)
  • Input: Any image — automatically resized to 32×32 RGB
  • Output: Softmax probabilities over the 10 classes
  • Best for: Single, centred, everyday objects. Photos of unrelated subjects (people, food, buildings) will still be forced into one of the 10 categories.