Skip to content

README.md

Latest commit

 

History

History
36 lines (21 loc) · 1.21 KB

README.md

File metadata and controls

36 lines (21 loc) · 1.21 KB

Datature Tensorflow Instance Segmentation Model Guide

This guide is for carrying out predictions using exported Datature Tensorflow Instance Segmentation models.

Requirements

To use GPU, the CUDA requirement is >= 11.2.

Install the required packages using:

pip3 install -r requirements.txt

Making Predictions

The predict.py file can be run as follows:

python3 predict.py \
    -i input_folder_path \
    -o output_folder_path \
    -m model_path \
    -l label_map_path \
    -t threshold

input_folder_path refers to the path to the folder where the images for prediction are stored.

output_folder_path refers to the path to the folder where the output images after prediction are to be stored. Do note that the output image names will be the same as the input image names so this should not be the same folder as the input folder.

model_path refers to the path to saved_model (not the saved_model directory)

label_map_path refers to the path to the label map file (not the label map file directory)

threshold refers to the threshold value in range (0.0, 1.0) for the prediction score. Only predictions with scores above the threshold value will be shown on the output image.