ayase 0.1.0

Modular media quality metrics toolkit

Ayase

Modular media quality metrics toolkit.

Overview

• 225 quality metrics across visual, temporal, audio, perceptual, and safety categories.
• Modular pipeline - modules compute raw values, downstream apps decide what to do with them.
• CLI and Python API.
• Profile-based pipeline configuration.

Installation

Core (no ML models, metadata/structural checks only):

pip install ayase

With ML-based quality metrics:

pip install ayase[ml]           # Everything
pip install ayase[v-perceptual] # CLIP, LPIPS, open-clip, timm
pip install ayase[v-iqa]        # PyIQA, BRISQUE, NIQE, scikit-video
pip install ayase[v-motion]     # RAFT optical flow, decord
pip install ayase[v-ocr]        # PaddleOCR text recognition
pip install ayase[v-face]       # MediaPipe face detection
pip install ayase[v-audio]      # librosa audio analysis

Development:

pip install ayase[dev]          # pytest, black, ruff, mypy
pip install ayase[all]          # Everything including dev + TUI

See MODELS.md for the complete inventory of all pretrained weights used by every module. Models are downloaded and cached automatically on first use via HuggingFace Hub, Torch Hub, and PyIQA.

Metrics

#	Metric	Title	Tested
1	blur_score	Blur Score	Yes
2	compression_score	Compression Score	Yes
3	aesthetic_score	Aesthetic Score	Yes
4	clip_score	CLIP Score	Yes
5	brightness	Brightness	Yes
6	contrast	Contrast	Yes
7	saturation	Saturation	Yes
8	fast_vqa_score	Fast VQA Score	Yes
9	motion_score	Motion Score	Yes
10	camera_motion_score	Camera Motion Score	Yes
11	temporal_consistency	Temporal Consistency	Yes
12	technical_score	Technical Score	Yes
13	noise_score	Noise Score	Yes
14	artifacts_score	Artifacts Score	Yes
15	watermark_probability	Watermark Probability	Yes
16	ocr_area_ratio	OCR Area Ratio	Yes
17	face_count	Face Count	Yes
18	nsfw_score	NSFW Score	Yes
19	audio_quality_score	Audio Quality Score	Yes
20	perceptual_hash	Perceptual Hash	Yes
21	depth_score	Depth Score	Yes
22	auto_caption	Auto Caption	Yes
23	vqa_a_score	VQA A Score	Yes
24	vqa_t_score	VQA T Score	Yes
25	is_score	IS Score	Yes
26	sd_score	SD Score	Yes
27	gradient_detail	Gradient Detail	Yes
28	blip_bleu	BLIP BLEU	Yes
29	detection_score	Detection Score	Yes
30	count_score	Count Score	Yes
31	color_score	Color Score	Yes
32	celebrity_id_score	Celebrity ID Score	Yes
33	ocr_score	OCR Score	Yes
34	ocr_fidelity	OCR Fidelity	Yes
35	i2v_clip	I2V CLIP	Yes
36	i2v_dino	I2V DINO	Yes
37	i2v_lpips	I2V LPIPS	Yes
38	i2v_quality	I2V Quality	Yes
39	action_score	Action Score	Yes
40	action_confidence	Action Confidence	Yes
41	flow_score	Flow Score	Yes
42	motion_ac_score	Motion AC Score	Yes
43	warping_error	Warping Error	Yes
44	clip_temp	CLIP Temporal	Yes
45	face_consistency	Face Consistency	Yes
46	psnr	PSNR	Yes
47	ssim	SSIM	Yes
48	lpips	LPIPS	Yes
49	spectral_entropy	Spectral Entropy	Yes
50	spectral_rank	Spectral Rank	Yes
51	fvd	FVD	Yes
52	kvd	KVD	Yes
53	fvmd	FVMD	Yes
54	vmaf	VMAF	Yes
55	ms_ssim	MS-SSIM	Yes
56	vif	VIF	Yes
57	niqe	NIQE	Yes
58	t2v_score	T2V Score	Yes
59	t2v_alignment	T2V Alignment	Yes
60	t2v_quality	T2V Quality	Yes
61	dynamics_range	Dynamics Range	Yes
62	dynamics_controllability	Dynamics Controllability	Yes
63	scene_complexity	Scene Complexity	Yes
64	compression_artifacts	Compression Artifacts	Yes
65	naturalness_score	Naturalness Score	Yes
66	video_memorability	Video Memorability	Yes
67	usability_rate	Usability Rate	Yes
68	confidence_score	Confidence Score	Yes
69	human_preference_score	Human Preference Score	Yes
70	engagement_score	Engagement Score	Yes
71	usability_score	Usability Score	Yes
72	hdr_quality	HDR Quality	Yes
73	sdr_quality	SDR Quality	Yes
74	temporal_information	Temporal Information	Yes
75	spatial_information	Spatial Information	Yes
76	flicker_score	Flicker Score	Yes
77	judder_score	Judder Score	Yes
78	stutter_score	Stutter Score	Yes
79	dists	DISTS	Yes
80	fsim	FSIM	Yes
81	gmsd	GMSD	Yes
82	vsi_score	VSI Score	Yes
83	brisque	BRISQUE	Yes
84	pesq_score	PESQ Score	Yes
85	av_sync_offset	A/V Sync Offset	Yes
86	dover_score	DOVER Score	Yes
87	dover_technical	DOVER Technical	Yes
88	dover_aesthetic	DOVER Aesthetic	Yes
89	topiq_score	TOPIQ Score	Yes
90	liqe_score	LIQE Score	Yes
91	clip_iqa_score	CLIP-IQA Score	Yes
92	color_grading_score	Color Grading Score	Yes
93	white_balance_score	White Balance Score	Yes
94	exposure_consistency	Exposure Consistency	Yes
95	focus_quality	Focus Quality	Yes
96	banding_severity	Banding Severity	Yes
97	qalign_quality	Q-Align Quality	Yes
98	qalign_aesthetic	Q-Align Aesthetic	Yes
99	face_quality_score	Face Quality Score	Yes
100	face_identity_consistency	Face Identity Consistency	Yes
101	face_expression_smoothness	Face Expression Smoothness	Yes
102	face_landmark_jitter	Face Landmark Jitter	Yes
103	object_permanence_score	Object Permanence Score	Yes
104	semantic_consistency	Semantic Consistency	Yes
105	depth_temporal_consistency	Depth Temporal Consistency	Yes
106	subject_consistency	Subject Consistency	Yes
107	background_consistency	Background Consistency	Yes
108	motion_smoothness	Motion Smoothness	Yes
109	codec_efficiency	Codec Efficiency	Yes
110	gop_quality	GOP Quality	Yes
111	codec_artifacts	Codec Artifacts	Yes
112	deepfake_probability	Deepfake Probability	Yes
113	ai_generated_probability	AI-Generated Probability	Yes
114	harmful_content_score	Harmful Content Score	Yes
115	watermark_strength	Watermark Strength	Yes
116	bias_score	Bias Score	Yes
117	depth_quality	Depth Quality	Yes
118	multiview_consistency	Multiview Consistency	Yes
119	stereo_comfort_score	Stereo Comfort Score	Yes
120	musiq_score	MUSIQ Score	Yes
121	contrique_score	CONTRIQUE Score	Yes
122	mdtvsfa_score	MDTVSFA Score	Yes
123	nima_score	NIMA Score	Yes
124	dbcnn_score	DBCNN Score	Yes
125	wadiqam_score	WaDIQaM Score	Yes
126	maniqa_score	MANIQA Score	Yes
127	arniqa_score	ARNIQA Score	Yes
128	qualiclip_score	QualiCLIP Score	Yes
129	pieapp	PieAPP	Yes
130	cw_ssim	CW-SSIM	Yes
131	nlpd	NLPD	Yes
132	mad	MAD	Yes
133	ahiq	AHIQ	Yes
134	topiq_fr	TOPIQ-FR	Yes
135	dreamsim	DreamSim	Yes
136	cover_score	Cover Score	Yes
137	cover_technical	Cover Technical	Yes
138	cover_aesthetic	Cover Aesthetic	Yes
139	cover_semantic	Cover Semantic	Yes
140	vqa_score_alignment	VQAScore Alignment	Yes
141	videoscore_visual	VideoScore Visual	Yes
142	videoscore_temporal	VideoScore Temporal	Yes
143	videoscore_dynamic	VideoScore Dynamic	Yes
144	videoscore_alignment	VideoScore Alignment	Yes
145	videoscore_factual	VideoScore Factual	Yes
146	face_iqa_score	Face IQA Score	Yes
147	scene_stability	Scene Stability	Yes
148	avg_scene_duration	Average Scene Duration	Yes
149	raft_motion_score	RAFT Motion Score	Yes
150	ram_tags	RAM Tags	Yes
151	depth_anything_score	Depth Anything Score	Yes
152	depth_anything_consistency	Depth Anything Consistency	Yes
153	video_type	Video Type	Yes
154	video_type_confidence	Video Type Confidence	Yes
155	jedi	JEDi	Yes
156	trajan_score	TRAJAN Score	Yes
157	promptiqa_score	PromptIQA Score	Yes
158	aigv_static	AIGV Static	Yes
159	aigv_temporal	AIGV Temporal	Yes
160	aigv_dynamic	AIGV Dynamic	Yes
161	aigv_alignment	AIGV Alignment	Yes
162	video_reward_score	Video Reward Score	Yes
163	text_overlay_score	Text Overlay Score	Yes
164	ptlflow_motion_score	PTLFlow Motion Score	Yes
165	qcn_score	QCN Score	Yes
166	finevq_score	FineVQ Score	Yes
167	kvq_score	KVQ Score	Yes
168	rqvqa_score	RQ-VQA Score	Yes
169	videval_score	VIDEVAL Score	Yes
170	tlvqm_score	TLVQM Score	Yes
171	funque_score	FUNQUE Score	Yes
172	movie_score	MOVIE Score	Yes
173	st_greed_score	ST-GREED Score	Yes
174	c3dvqa_score	C3DVQA Score	Yes
175	flolpips	FloLPIPS	Yes
176	hdr_vqm	HDR-VQM	Yes
177	st_lpips	ST-LPIPS	Yes
178	camera_jitter_score	Camera Jitter Score	Yes
179	jump_cut_score	Jump Cut Score	Yes
180	playback_speed_score	Playback Speed Score	Yes
181	flow_coherence	Flow Coherence	Yes
182	letterbox_ratio	Letterbox Ratio	Yes
183	vtss	VTSS	Yes
184	cnniqa_score	CNNIQA Score	Yes
185	hyperiqa_score	HyperIQA Score	Yes
186	paq2piq_score	PaQ2PiQ Score	Yes
187	tres_score	TReS Score	Yes
188	unique_score	Unique Score	Yes
189	laion_aesthetic	LAION Aesthetic	Yes
190	compare2score	Compare2Score	Yes
191	afine_score	A-FINE Score	Yes
192	ckdn_score	CKDN Score	Yes
193	deepwsd_score	DeepWSD Score	Yes
194	ssimulacra2	SSIMULACRA2	Yes
195	butteraugli	Butteraugli	Yes
196	flip_score	Flip Score	Yes
197	vmaf_neg	VMAF-NEG	Yes
198	ilniqe	ILNIQE	Yes
199	nrqm	NRQM	Yes
200	pi_score	PI Score	Yes
201	piqe	PIQE	Yes
202	maclip_score	MACLIP Score	Yes
203	dmm	DMM	Yes
204	wadiqam_fr	WaDIQaM-FR	Yes
205	ssimc	SSIM-C	Yes
206	cambi	CAMBI	Yes
207	xpsnr	XPSNR	Yes
208	vmaf_phone	VMAF-Phone	Yes
209	vmaf_4k	VMAF-4K	Yes
210	visqol	ViSQOL	Yes
211	dnsmos_overall	DNSMOS Overall	Yes
212	dnsmos_sig	DNSMOS Signal	Yes
213	dnsmos_bak	DNSMOS Background	Yes
214	pu_psnr	PU-PSNR	Yes
215	pu_ssim	PU-SSIM	Yes
216	max_fall	Max FALL	Yes
217	max_cll	Max CLL	Yes
218	hdr_vdp	HDR-VDP	Yes
219	delta_ictcp	Delta ICtCp	Yes
220	ciede2000	CIEDE2000	Yes
221	psnr_hvs	PSNR-HVS	Yes
222	psnr_hvs_m	PSNR-HVS-M	Yes
223	cgvqm	CGVQM	Yes
224	strred	STRRED	Yes
225	p1203_mos	P.1203 MOS	Yes

Quick Start

CLI

# Scan a dataset and get a report
ayase scan ./my_dataset

# Scan with specific modules
ayase scan ./my_dataset --modules metadata,basic_quality,motion

# List all available modules
ayase modules list

# Check which modules can be loaded (dependencies installed)
ayase modules check

# Filter dataset by quality score
ayase filter ./my_dataset --min-score 70 --output ./filtered

Python API (recommended)

from ayase import AyasePipeline

ayase = AyasePipeline(modules=["basic"])
results = ayase.run("./my_dataset")

for path, sample in results.items():
    if sample.quality_metrics:
        print(f"{sample.path.name}: technical={sample.quality_metrics.technical_score}")

print(f"Total: {ayase.stats.total_samples}, Valid: {ayase.stats.valid_samples}")
ayase.export("report.json")

AyasePipeline accepts three ways to configure modules:

# By module names
ayase = AyasePipeline(modules=["metadata", "basic_quality", "motion"])

# By profile dict
ayase = AyasePipeline(profile={
    "name": "my_check",
    "modules": ["basic", "aesthetic"],
    "module_config": {
        "aesthetic": {"model_name": "openai/clip-vit-large-patch14"},
    },
})

# By profile file
ayase = AyasePipeline(profile="my_profile.toml")

# With custom config
from ayase.config import AyaseConfig
ayase = AyasePipeline(config=AyaseConfig(general={"parallel_jobs": 16}), modules=["basic"])

Low-level Pipeline API

import asyncio
from pathlib import Path
from ayase.pipeline import Pipeline, ModuleRegistry
from ayase.scanner import scan_dataset

ModuleRegistry.discover_modules()
module_names = ["metadata", "basic_quality", "semantic_alignment"]
modules = [ModuleRegistry.get_module(n)() for n in module_names]

pipeline = Pipeline(modules)
pipeline.start()

samples = scan_dataset(Path("./my_dataset"), recursive=True)
for sample in samples:
    processed = asyncio.run(pipeline.process_sample(sample))

pipeline.stop()
pipeline.export_report("report.json", format="json")

Profile-based pipelines

from ayase import load_profile, instantiate_profile_modules

profile = load_profile("my_profile.toml")
modules = instantiate_profile_modules(profile)
# modules is a list of PipelineModule instances ready for Pipeline()

Configuration

Create ayase.toml in your project root:

[general]
parallel_jobs = 8
cache_enabled = true

[quality]
enable_blur_detection = true
blur_threshold = 100.0

[pipeline]
dataset_path = "./my_dataset"
modules = ["metadata", "basic_quality", "motion"]
plugin_folders = ["plugins"]

[output]
default_format = "markdown"
artifacts_dir = "reports"
artifacts_format = "json"

[filter]
default_mode = "list"
min_score_threshold = 60

Ayase looks for config in: ./ayase.toml -> ~/.config/ayase/config.toml -> built-in defaults.

Writing Plugins

Create a .py file in your plugins/ folder:

from ayase.models import Sample, ValidationIssue, ValidationSeverity
from ayase.pipeline import PipelineModule

class MyCustomCheck(PipelineModule):
    name = "my_check"
    description = "Custom quality check"
    default_config = {"threshold": 0.5}

    def process(self, sample: Sample) -> Sample:
        # Your logic here
        if some_score < self.config["threshold"]:
            sample.validation_issues.append(
                ValidationIssue(
                    severity=ValidationSeverity.WARNING,
                    message="Quality below threshold",
                )
            )
        return sample

Then run:

ayase scan ./data --modules metadata,my_check

Development

git clone <repo-url>
cd ayase
pip install -e ".[dev]"

# Run tests
pytest

# Lint and format
ruff check src/ tests/
black src/ tests/

# Type check
mypy src/ayase

License

MIT -- see LICENSE.