4.1.1 A quick review of FV input and output interface

Overall

Before	After
load dataset from local load weights from local validating save result at local	load dataset from os.environ["INPUT_PATH"] # step 2 load model weights from os.environ["WEIGHT_PATH"] # step 2 validating Prepare an output dictionary # step 1 save the output dictionary to os.environ["LOCAL_MODEL_PATH"] # step 2

Input interface

INPUT_PATH with examples

Before	After
# import cv2 img = cv2.imread("./dataset/img1.jpg")	# import cv2 img = cv2.imread(os.environ["INPUT_PATH"]+"/dataset/img1.jpg")
# import pandas as pd df = pd.read_csv("./data.csv")	# import pandas as pd df = pd.read_csv(os.environ["INPUT_PATH"]+"/data.csv")

WEIGHT_PATH: Load pre-trained weights from the path into the model.

OUTPUT_PATH: Save the prepared dictionary as $OUTPUT_PATH/result.json. Remember to import json in advance.

Output interface

metadata: dict
- dataSize: int # MUST. It is the number of the validation data.
results: dict
The categories are all OPTIONAL. If len(tables) is 3, you will see 3 tables on the dashboard. Each instance in a category is passed by a dict. e.g. tables["title"] means the title of the table.
- tables: list[dict] # It's 1-D table, so input 1-D column name and value.
  - title: str
  - labels: list[str]
  - values: list[num]
- bars: list[dict]
  - title: str
  - labels: list[str] # e.g. names of M categories
  - y-axis: str
  - values: list[num] # e.g. values of M categories
- heatmaps: list[dict]
  - title: str
  - x-axis: str
  - y-axis: str
  - x-labels: list[str]
  - y-labels: list[str]
  - values: list[list[num]]
- plots: list[dict]
  - title: str
  - x-axis: str
  - y-axis: str
  - x-values: list[list[str]] # e.g. M*N means M categories with each N time series points
  - y-values: list[list[str]] # e.g. M*N means M categories with each N time series points
  - labels: list[str] # e.g. Must be M categories as above
- images: list[dict]
  - title: str
  - filename: str # The image name saved in OUTPUT_PATH