Skip to content

Latest commit

 

History

History
9 lines (7 loc) · 947 Bytes

gse_train_README.md

File metadata and controls

9 lines (7 loc) · 947 Bytes

The Group Size Estimator (GSE) Training Blueprint provides an end-to-end framework for finetuning object counter models on custom image datasets. This Bleurprint lets the user finetune a pre-trained YOLOv5 model on custom data (can be downloaded via the S3 Connector) and then use it to perform inference and batch predictions. Complete the following steps to run this blueprint:

  1. Click the Use Blueprint button.
  2. Select a suitable Compute Template to run the blueprint libraries.
  3. Enter and/or edit input arguments to the blueprint libraries and save the changes. Refer to the documentation pages of libraries for more details.
  4. Click the Run button to launch the end-to-end framework.
  5. Once the inference endpoint is ready, use either the Try it Live feature or the Integration panel to integrate the endpoint into your code.

Click here for more information on this Blueprint.