EfficientNet-Lite4 — ONNX

ONNX export of EfficientNet-Lite4, a lightweight and efficient image classification model optimized for mobile/edge deployment. Trained on ImageNet with 1000-class output.

Mirrored for use with inference4j, an inference-only AI library for Java.

Original Source

• Repository: ONNX
• License: apache-2.0

Usage with inference4j

try (EfficientNet model = EfficientNet.fromPretrained("models/efficientnet-lite4")) {
    List<Classification> results = model.classify(Path.of("cat.jpg"));
    results.forEach(c -> System.out.printf("%s: %.2f%%%n", c.label(), c.score() * 100));
}

Model Details

Property	Value
Architecture	EfficientNet-Lite4 (compound-scaled CNN)
Task	Image classification (ImageNet 1000 classes)
Input	[batch, 224, 224, 3] — RGB, pixel values 0-255
Output	[batch, 1000] — class probabilities
ONNX opset	11
Original framework	TensorFlow Lite → ONNX

License

This model is licensed under the Apache License 2.0. Original model from ONNX.