Adding ConfusionMatrixEvaluator

New evaluator added to compute confusion matrix (https://en.wikipedia.org/wiki/Confusion_matrix)

What's Changed

• Make pycococap package optional dependency by @shonohs in #41
• adding confusionmatrix evaluator by @gegeo0 in #40

New Contributors

• @shonohs made their first contribution in #41

Full Changelog: 0.2.13...0.2.14

Fix MAP@K bug

Issue fix for MAP@K. Top-K predictions were not ordered which resulted to incorrect computation of map@k

Recall@k, Precision@k, Mean Average Precision@k, Precision-Recall Curve for image retrieval

Added the following metrics relevant to image retrieval:
Recall@k, Precision@k, Mean Average Precision@k, Precision-Recall Curve

Fix mAP evaluator bug

Fix bug: when category ids are not continuous and the prediction / label is tuple instead of list, vision-evaluation throws an exception when attempting to make category ids continuous.

Support indices predictions for prediction filter

Update:

1. Adding support for indices predictions for prediction filter
2. Adding MeanLpError evaluator

fix dependencies and home page link

0.2.9

fix dependencies and home page link (#31)

Fix AP for multilabel+macro

0.2.8

fix ap calculation under macro for multilabel classification (#30)

Implement group-wise evaluator

0.2.7

Added GroupWiseEvaluator  (#29)

Improve efficiency of top1/5 accuracy and AP macro

improve efficiency/memory of ap macro and topk acc (#25)

improve efficiency/memory of ap macro and topk acc

Support image matting

0.2.5

Support image matting (#23)