hailo_model_zoo

Why we included this project

Teams building computer-vision products on Hailo's edge accelerators will find this the natural place to start. The zoo ships pre-trained networks in ONNX and TensorFlow formats, plus pre-compiled HEF binaries that run directly on Hailo devices, with models for classification, object detection, segmentation, and similar tasks across the Hailo-8, Hailo-10, and Hailo-15 lines. It also includes the tooling to measure full-precision accuracy, check quantized accuracy through the Hailo emulator, and evaluate on real hardware before producing a deployable binary. If you need to adapt a model to your own data, the retraining instructions walk through fine-tuning a zoo network on a custom dataset. That makes it a practical way to judge whether Hailo hardware.