Skip to content

Model Deployment

To successfully deploy your model, we require the following components:

  1. Model Overview:
    • Please provide a concise description of the model, outlining the specific problem it addresses and its approach to solving that problem.
  2. Base Model or Deep Learning Architecture
    • If your model is built upon an existing architecture or framework, such as Yolo family, Detectron2, ResNet, etc., please specify the architecture.
    • If the model was designed from scratch, include the architecture details in PyTorch, TensorFlow, or ONNX format.
  3. Analytics/Post-processor code
    • please provide the code If there is any analytics or post-processor algorithms are needed on top of the main model.
  4. Desired Output Format
    • Specify the required columns or data in the output CSV. This will help ensure that the results from your model align with our needs.
  5. Annotation Preparation
    • Clarify how you plan to prepare annotations for detections or AI-Clip results. Providing details about the annotation process will help streamline integration.
  6. Sample Output
    • Including a sample output will offer clarity on the expected format and assist in verifying the results.
  7. Dependencies
    • State the versions of PyTorch/TensorFlow and any other third-party libraries that your model depends on.
    • Additionally, provide information about the CUDA version required for GPU acceleration.
  8. Deployment Script
    • Preferably, supply a simple script that allows us to execute your model on our end. This will serve as a validation step, confirming that the model functions as intended in the deployment environment.

By providing these details, we can ensure a smooth deployment process and verify that your model aligns with our requirements.