Animals-10 · CNN from scratch

Classifies a photo into one of 10 animals: butterfly, cat, chicken, cow, dog, elephant, horse, sheep, spider, squirrel.

A small convolutional network trained from scratch (no pre-trained weights): 6 conv layers, batch norm, dropout and global average pooling, ~0.76 M parameters, 90x90 input. It is the baseline the transfer-learning models are compared against.

Metric (held-out test, n=2618) Value
Accuracy 0.885
F1 macro 0.875
Precision (macro) 0.876
Recall (macro) 0.875

Confusion matrix

Data: Animals-10, 26,177 photos, stratified split 80/10/10 (train / validation for early stopping / untouched test). The four models of the project share exactly the same split, so their scores are directly comparable.

Training: input 90x90, 41 epochs, on NVIDIA GeForce RTX 5060 Laptop GPU (9.0 min). Code: https://github.com/JTiiz/CNN-project (training/).

Usage

import os
os.environ["KERAS_BACKEND"] = "torch"  # the backend the model was trained and tested with

import json, keras, numpy as np
from huggingface_hub import snapshot_download
from PIL import Image

path = snapshot_download("jramduran/animals10-cnn-scratch")
info = json.load(open(f"{path}/model_info.json"))
model = keras.saving.load_model(f"{path}/model.keras")

size = info["input_size"]
img = Image.open("cat.jpg").convert("RGB").resize((size, size))
x = np.asarray(img, dtype="float32")[None]          # raw pixels 0-255, no normalisation
probs = model.predict(x, verbose=0)[0]
print(info["labels"][probs.argmax()], probs.max())
Downloads last month
3
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Dataset used to train jramduran/animals10-cnn-scratch