February 12, 2021 (Python SDK)

We are committed to your success, that's why we are always improving our platform. You can now integrate training and prediction calls within the Python pipeline.

Run model training

To run a model training:

sa.run_training(
  project = ["city_batch1","city_batch2"], 
  base_model = "Instance Segmentation (trained on COCO)", 
  model_name = "Cityscapes Segmentation Model", 
  model_description = "Model Finetuned on Cityscapes", 
  task = "Instance Segmentation for Vector Projects", 
  hyperparameters = {"base_lr": 0.02, "images_per_batch": 8},
  log = True)

Adjust the default values of the training hyperparameters based on the following table:

Parameter	Description	Default Value
instance_type	The type of the devices (GPUs) available to train your model. You have two options: One GPU of type T4 with 16GB memory or 4 GPUs of the same type	1 x T4 16 GB
num_epochs	The number of times the dataset undergoes training. The epoch count ranges from 10 to 200.	12
images_per_batch	Number of images fed to network per training iteration per GPU	4
base_lr	The model’s initial learning step. The learning rate ranges from 0 to 1.	0.02
batch_per_image	The number of regions of interest (ROI) per image. The ROI ranges from 2 to 512 and is a power of 2. The smaller the ROI, the faster the learning rate. Image (ROIs) per batch works only with Instance segmentation.	512
evaluation_period	The number of epochs to run evaluation periodically. If the epoch count is 10 and the evaluation period is 2, it means that the dataset will undergo an evaluation every 2 epochs. The user will be reported of the best model that resulted from the evaluation period.	12
dataset_split_ratio	The number of images that need to be trained to the number of images that need to be tested. The train/test split ratio ranges from 0 to 100.	80
gamma	Gamma is needed for the learning rate. Its value ranges from 0 to 1, and it will be multiplied with the learning rate.	0.5
steps	List of epoch counts to decrease learning rate gamma times	[ 5 ]

Run smart prediction

To run smart prediction on a model:

project_name = "city_batch3"

images = sa.search_images(
  project = project_name,
	annotation_status = "NotStarted")

succeeded, failed = sa.run_prediction(
  project = project_name, 
  images_list = images,
  model = "Cityscapes Segmentation Mode")

This function logs the progress bar which shows the number of successful predicted images.

Run smart segmentation

To run smart segmentation on a model:

succeeded, failed = sa.run_segmentation(
  project = project_name, 
  images_list = images,
  model = "generic")

There are two segmentation model available: generic and autonomous. Choose the option that best suits your use case.

Run model training

Run smart prediction

Run smart segmentation

Other changes and improvements