Metrics

31 public types in this namespace, organized by kind.

Models & Types (27)

Type Summary
AestheticScore<T> Aesthetic Score metric using CLIP for evaluating image aesthetics.
ApproximatePerceptualSpeechQuality<T> A PESQ-like speech quality approximation.
CLIPScore<T> CLIPScore metric for evaluating text-image alignment and image quality.
ChamferDistance<T> Chamfer Distance metric for 3D point cloud comparison.
CharacterErrorRate Character Error Rate (CER) metric for speech recognition and OCR evaluation.
EarthMoversDistance<T> Earth Mover’s Distance (EMD) / Wasserstein Distance for point cloud comparison.
FScore<T> F-Score metric for 3D reconstruction evaluation.
FrechetInceptionDistance<T> Fréchet Inception Distance (FID) - A metric for evaluating the quality of generated images.
FrechetVideoDistance<T> Fréchet Video Distance (FVD) - A metric for evaluating the quality of generated videos.
InceptionScore<T> Inception Score (IS) - A metric for evaluating the quality and diversity of generated images.
IoU3D<T> 3D Intersection over Union (3D IoU) for voxel and bounding box evaluation.
KernelInceptionDistance<T> Kernel Inception Distance (KID) - A metric for evaluating the quality of generated images.
MeanIntersectionOverUnion<T> Mean Intersection over Union (mIoU) metric for segmentation tasks.
ObjectDetectionMetrics<T> COCO-style evaluation metrics for object detection: Average Precision (AP), mean Average Precision (mAP) and the underlying precision-recall curve.
OverallAccuracy<T> Overall Accuracy metric for classification and segmentation.
PeakSignalToNoiseRatio<T> Peak Signal-to-Noise Ratio (PSNR) metric for image quality assessment.
ScaleInvariantSignalToDistortionRatio<T> Scale-Invariant Signal-to-Distortion Ratio (SI-SDR) metric for source separation evaluation.
ShortTimeObjectiveIntelligibility<T> Short-Time Objective Intelligibility (STOI) metric for speech intelligibility assessment.
SignalToNoiseRatio<T> Signal-to-Noise Ratio (SNR) metric for audio quality assessment.
StructuralSimilarity<T> Structural Similarity Index Measure (SSIM) for image quality assessment.
TemporalConsistency<T> Temporal Consistency metric for evaluating video smoothness and coherence.
TextDetectionMetrics<T> ICDAR-style evaluation metrics for text detection: polygon IoU, and the precision, recall and H-mean triple that the ICDAR Robust Reading competitions report.
VideoPSNR<T> Video Peak Signal-to-Noise Ratio (VPSNR) - Frame-averaged PSNR for video quality.
VideoQualityIndex<T> Video Quality Index (VQI) - A composite metric combining multiple video quality aspects.
VideoQualityResult<T> Results from comprehensive video quality evaluation.
VideoSSIM<T> Video Structural Similarity (VSSIM) - Frame-averaged SSIM for video quality.
WordErrorRate Word Error Rate (WER) metric for speech recognition evaluation.

Enums (1)

Type Summary
FrameSamplingStrategy Frame sampling strategies for video feature extraction.

Helpers & Utilities (3)

Type Summary
LanguageModelMetrics<T> Static helpers for evaluating language-model / classification predictions over a vocabulary: perplexity (how surprised the model is by the true tokens) and top-k accuracy (how often the true token is among the model’s k most likely…
RankingMetrics<T> Static helpers for evaluating the quality of a ranking, most notably Normalized Discounted Cumulative Gain (NDCG@k).
TextRecognitionMetrics Evaluation metrics for text recognition (OCR): Character Error Rate, Word Error Rate, normalized edit distance and exact-match accuracy.