Skip to content

Params

Evaluation parameters. Created automatically by COCOeval, but can be modified before calling evaluate().

ev = COCOeval(coco_gt, coco_dt, "bbox")
ev.params.max_dets = [1, 10, 100]
ev.params.area_rng = [[0, 10000000000]]
ev.params.area_rng_lbl = ["all"]

let mut ev = COCOeval::new(coco_gt, coco_dt, IouType::Bbox);
ev.params.max_dets = vec![1, 10, 100];
ev.params.area_rng = vec![[0.0, 1e10]];
ev.params.area_rng_lbl = vec!["all".to_string()];

Constructor

Params(iou_type: str = "bbox")

Params::new(iou_type: IouType) -> Self

You rarely need to construct Params directly — COCOeval creates one automatically.

Properties

iou_type

Evaluation type.

Python Rust
Type str IouType
Default "bbox" IouType::Bbox
Values "bbox", "segm", "keypoints" Bbox, Segm, Keypoints

camelCase alias

Also available as iouType in Python.

img_ids

Image IDs to evaluate. Empty list means all images.

Python Rust
Type list[int] Vec<u64>
Default [] vec![]

camelCase alias

Also available as imgIds in Python.

cat_ids

Category IDs to evaluate. Empty list means all categories.

Python Rust
Type list[int] Vec<u64>
Default [] vec![]

camelCase alias

Also available as catIds in Python.

iou_thrs

IoU thresholds for evaluation.

Python Rust
Type list[float] Vec<f64>
Default [0.5, 0.55, 0.6, ..., 0.95] (10 values) Same

camelCase alias

Also available as iouThrs in Python.

rec_thrs

Recall thresholds for precision interpolation.

Python Rust
Type list[float] Vec<f64>
Default [0.0, 0.01, 0.02, ..., 1.0] (101 values) Same

camelCase alias

Also available as recThrs in Python.

max_dets

Maximum detections per image. The summary metrics report results at each of these thresholds.

Python Rust
Type list[int] Vec<usize>
Default (bbox/segm) [1, 10, 100] Same
Default (keypoints) [20] Same

camelCase alias

Also available as maxDets in Python.

area_rng

Area ranges for size-based evaluation. Each range is [min_area, max_area] in pixels.

Python Rust
Type list[list[float]] Vec<[f64; 2]>
Default (bbox/segm) [[0, 1e10], [0, 1024], [1024, 9216], [9216, 1e10]] Same
Default (keypoints) [[0, 1e10], [1024, 9216], [9216, 1e10]] Same

The defaults correspond to: all, small (0-32²), medium (32²-96²), large (96²+). Keypoints skip the small range.

camelCase alias

Also available as areaRng in Python.

area_rng_lbl

Labels for the area ranges.

Python Rust
Type list[str] Vec<String>
Default (bbox/segm) ["all", "small", "medium", "large"] Same
Default (keypoints) ["all", "medium", "large"] Same

camelCase alias

Also available as areaRngLbl in Python.

use_cats

Whether to evaluate per-category. When False, all detections and ground truth annotations are pooled regardless of category label.

Python Rust
Type bool bool
Default True true

camelCase alias

Also available as useCats in Python.

kpt_oks_sigmas

Per-keypoint OKS sigma values. Controls how strictly each keypoint is evaluated — higher sigma means more tolerance.

Python Rust
Type list[float] Vec<f64>
Default 17 COCO keypoint sigmas Same

Default values (nose, eyes, ears, shoulders, elbows, wrists, hips, knees, ankles):

[0.026, 0.025, 0.025, 0.035, 0.035, 0.079, 0.079, 0.072, 0.072,
 0.062, 0.062, 0.107, 0.107, 0.087, 0.087, 0.089, 0.089]

camelCase alias

Also available as kptOksSigmas in Python.