Introducing IConvEx-1
IConvEx-1 is an experimental closed-set recognition image classification Convolutional Neural Network (CNN), developed and trained by Coombs Technologies.
Specifications
This model is trained to classify images into the following 20 classes:
- Aeroplane
- Bicycle
- Bird
- Boat
- Book
- Car
- Cat
- Computer
- Crocodile
- Dog
- Flower
- Guitar
- Hamster
- House
- Human
- Mobile Phone
- Motorcycle
- Snake
- Spider
- Tree
The input image size is 64x64 pixels with 3 RGB channels.
Usage
import onnxruntime as ort
import numpy as np
from PIL import Image
session = ort.InferenceSession("model.onnx")
img = Image.open("image.jpg").resize((64, 64))
img = np.array(img) / 255.0
img = img.transpose(2, 0, 1)[None].astype(np.float32)
outputs = session.run(None, { "input": img })
print(outputs)
Limitations
- This model was trained on 64x64 images and may perform poorly on higher resolution images without resizing.
- Performance may decrease for out-of-distribution data.
- This model is experimental and intended for research purposes.