Implementation:Obss Sahi Coco2fiftyone

Knowledge Sources	Obss_Sahi FiftyOne
Domains	Visualization, COCO, Evaluation
Last Updated	2026-02-08 12:00 GMT

Overview

CLI script that loads COCO datasets and detection results into FiftyOne for interactive visualization and evaluation.

Description

The coco2fiftyone script provides a command-line interface for importing COCO-formatted datasets and detection results into the FiftyOne visualization platform. It creates a FiftyOne dataset from COCO images and annotations, optionally adds one or more detection result JSON files as prediction labels, evaluates the first result set against ground truth (computing false positives), and launches the FiftyOne app sorted by false positive count for error analysis.

Usage

Run this script from the sahi CLI when you want to visually inspect detection results against COCO ground truth using FiftyOne, particularly for analyzing false positive patterns.

Code Reference

Source Location

• Repository: Obss_Sahi
• File: sahi/scripts/coco2fiftyone.py
• Lines: 1-81

Signature

def main(
    image_dir: str,
    dataset_json_path: str,
    *result_json_paths,
    iou_thresh: float = 0.5,
) -> None:
    """
    Args:
        image_dir (str): directory for coco images
        dataset_json_path (str): file path for the coco dataset json file
        result_json_paths (str): one or more paths for the coco result json file
        iou_thresh (float): iou threshold for coco evaluation
    """

Import

from sahi.scripts.coco2fiftyone import main

I/O Contract

Inputs

Name	Type	Required	Description
image_dir	str	Yes	Directory containing COCO images
dataset_json_path	str	Yes	Path to COCO annotation JSON file
result_json_paths	str (variadic)	No	One or more COCO result JSON paths
iou_thresh	float	No (default 0.5)	IOU threshold for evaluation

Outputs

Name	Type	Description
FiftyOne session	Interactive app	Browser-based visualization at localhost

Usage Examples

Visualize Dataset with Results

# Via CLI (using python-fire)
python -m sahi.scripts.coco2fiftyone \
    --image_dir /path/to/images \
    --dataset_json_path /path/to/annotations.json \
    /path/to/result1.json /path/to/result2.json \
    --iou_thresh 0.5

Programmatic Usage

from sahi.scripts.coco2fiftyone import main

main(
    image_dir="/data/coco/images",
    dataset_json_path="/data/coco/annotations.json",
    "/data/results/model_a.json",
    "/data/results/model_b.json",
    iou_thresh=0.5,
)

Related Pages

• Environment:Obss_Sahi_Python_Detection_Frameworks