Detection Categories
Many detections in Valossa Metadata include a categ field containing category tags. These tags classify detections into broad categories such as "fauna", "sport", or "content_compliance", making it easy to filter detections by theme.
How Categories Work
The categ field contains a tags array with one or more category identifier strings:
{
"t": "visual.context",
"label": "dog",
"categ": {
"tags": ["fauna", "pets"]
}
}
Category tags are available for most visual.context detections and some audio.context detections.
note
Detection categories describe individual concepts (e.g., a "dog" has the tags "fauna" and "pets"). This is different from video-level categories like topic.iab which describe what the entire video is about.
Non-Sensitive Tag Categories
These 27 categories cover general, non-sensitive content.
| Display Name | Tag Identifier | Description |
|---|---|---|
| Automotive | automotive | Cars, trucks, and motorcycles |
| Aviation | aviation | Airplanes and spacecraft |
| Boats and ships | boats_ships | Boats and ships |
| Brand or product | brand_product | Brands and products (excluding logos) |
| Buildings and architecture | buildings_architecture | Buildings and architectural details |
| Celebrations and holidays | celebrations_holidays | Weddings, graduation, Christmas, Thanksgiving, etc. |
| Children, family and play | children_family_play | Children, toys, board games, amusement parks |
| Computers and video games | computers_video_games | Computers and video games |
| Consumer electronics | consumer_electronics | Mobile devices, gadgets, cameras, TVs, appliances |
| Fashion and wear | fashion_wear | Clothing, shoes, accessories, jewelry, makeup |
| Animals | fauna | Animals |
| Plants and mushrooms | flora | Plants and mushrooms |
| Food or drink | food_drink | Foods, drinks, eating, and drinking |
| Graphics | graphics | Graphics that enrich media content |
| Home and garden | home_garden | Interior design, furniture, rooms, garden |
| Basic human actions | human_basic | Sitting, standing, sleeping, etc. |
| Human features and body parts | human_features | Human features and body parts (excluding emotions and compliance content) |
| Social situations and human life | human_life_social | Social activities, gambling, parades, students |
| Industrial | industrial | Machinery, power plants, cables |
| Lights and effects | lights_effects | Light-based media effects |
| Materials | materials | Concrete, wood, iron, etc. |
| Military equipment and people | military | Military staff, vehicles, aircraft, vessels |
| Music | music | Musical instruments, events, settings |
| Natural phenomena | natural_phenomena | Natural phenomena and events |
| Landscape and environment | nonlive_natural | Rock, mountain, sun, river, glacier |
| Other man-made objects | other_manmade_object | Objects not in any other specific category |
| Pets | pets | Dogs, cats, and their accessories |
| Place, location or scene | place_scene | Living room, stadium, road, etc. |
| Professions and work | professions_work | Humans at work (excluding athletes and military) |
| Public transport | public_transport | Bus, train, etc. |
| Sports | sport | Sports, sporting events, athletes |
| Sports equipment | sport_equipment | Sports equipment, protection, clothing |
| Sport locations | sport_locations | Swimming pool, stadium, etc. |
| Style | style | Image styles such as diagram or cartoon |
| Traffic, traffic areas and signs | traffic | Traffic stations, roads, signs, congestion |
| Travel destinations | travel_destinations | Famous landmarks around the world |
| Video structure | video_structure | Black frames, structural elements |
| Visual arts and crafts | visual_arts_crafts | Sculptures, paintings, handicrafts |
Sensitive Tag Categories
These 24 categories cover sensitive, potentially inappropriate, or compliance-related content. They are derived from visual, speech, and OCR detections.
| Display Name | Tag Identifier | Description |
|---|---|---|
| Accidents and destruction | accident | Accidents, explosions, conflicts, catastrophe aftermath |
| Act of violence | act_of_violence | Acts of violence that could injure a victim |
| Bad language | bad_language | Swearing and profanity |
| Bombs and explosions | bomb_explosion | Explosions and smoke |
| Content compliance | content_compliance | Master tag present on all Content Compliance concepts |
| Crime | crime | Criminal activity |
| Death | death | Death-related content |
| Explicit content | explicit_content | Highly sexual and violent content (always paired with more specific tags) |
| Guns and weapons | gun_weapon | Guns, knives, bows |
| Hate speech | hate_speech | Hate speech |
| Injury | injury | Blood, wounds, bruises |
| Military conflict | military_conflict | War, military conflict |
| Natural disasters and severe weather | natural_disaster_severe_weather | Floods, thunderstorms, etc. |
| Online piracy | online_piracy | Online piracy indicators |
| Religion | religion | Religious symbols, persons, places of worship |
| Revealing | revealing | Revealing clothing, visible body parts (navel, cleavage) |
| Sensitive social issues | sensitive_social_issues | Sensitive social issues |
| Sensual | sensual | Content hinting at sexuality (bikinis, kissing, etc.) |
| Sexual | sexual | Clearly sexual material and intimate body parts |
| Substance use | substance_use | Smoking, drugs, medicines, alcohol |
| Terrorism | terrorism | Terrorism |
| Threat of violence | threat_of_violence | Threats toward a person (e.g., aiming a gun) |
| Under 18 years | under_18_years | People under 18 years of age |
| Violence, injuries and threats | violence | Acts and signs of violence |
Content Compliance Filtering
The content_compliance tag serves as a master tag for all sensitive content. A detection with content_compliance will also have one or more specific sensitive tags (e.g., violence, sexual, substance_use).
Find All Content Compliance Detections
content_compliance = {}
for det_id, detection in metadata["detections"].items():
if "categ" in detection:
if "content_compliance" in detection["categ"]["tags"]:
content_compliance[det_id] = detection
Find Content Compliance Only in Visual Detections
content_compliance = {}
visual_ids = metadata["detection_groupings"]["by_detection_type"].get("visual.context", [])
for det_id in visual_ids:
detection = metadata["detections"][det_id]
if "categ" in detection:
if "content_compliance" in detection["categ"]["tags"]:
content_compliance[det_id] = detection
Filter by Specific Sensitive Category
violence_detections = {}
for det_id, detection in metadata["detections"].items():
if "categ" in detection:
tags = detection["categ"]["tags"]
if "violence" in tags or "act_of_violence" in tags:
violence_detections[det_id] = detection
Related Resources
- Content Moderation Guide -- Complete workflow for content compliance
- IAB Categories -- Video-level topic categorization