YOLO12-x is the extra-large, high-performance variant of the 12th-generation YOLO series, designed for extremely complex visual tasks and high-end computing environments. The model significantly expands both depth and width, utilizing advanced network architectures and self-attention mechanisms combined with the latest multi-scale feature fusion techniques, greatly enhancing detection of small, occluded, and complex-background objects.

YOLO12-x supports richer semantic extraction and contextual understanding, catering to demanding applications such as autonomous driving, intelligent surveillance, large-scale industrial inspection, and medical imaging analysis. Despite its larger parameter count and computational requirements, optimized inference strategies and distributed computing allow for relatively smooth real-time performance.

The model is compatible with mainstream inference frameworks like ONNX, TensorRT, and OpenVINO, and supports hardware acceleration across GPUs, TPUs, and AI accelerator cards. YOLO12-x is the top choice for visual AI developers and enterprises pursuing peak performance.

Source model

• Input shape: 1x3x640x640
• Number of parameters: 56.44M
• Model size: 225.99M
• Output shape: 1x84x8400

Source model repository: yolo12x

When the user has fine-tuned the source model, the model conversion process must be performed again.

Users can refer to either of the following two methods to complete the model conversion:

• Using AIMO for model conversion: Click Model Conversion Reference in the Performance Reference section on the right to view the conversion steps.

• Using Qualcomm QNN for model conversion: Please refer to the Qualcomm QNN Documentation.

The model performance benchmarks and example code provided by Model Farm are all implemented based on the APLUX AidLite SDK.

For models in .bin format, you can use either of the following two inference engines to run inference on Qualcomm chips:

Inference using APLUX AidLite: please refer to the APLUX AidLite Developer Documentation

Inference using Qualcomm QNN: Please refer to the Qualcomm QNN Documentation

Inference Example Code

The inference example code is implemented using the AidLite SDK.

Click Model & Code to download the model files and the inference code package. The file structure is as follows:

/model_farm_{model_name}_aidlite
    
    |__ models # folder where model files are stored

    |__ python # aidlite python model inference example

    |__ cpp # aidlite cpp model inference example

    |__ README.md