| AIBOX SP-500X | |
| Processador | Rockchip RK3568 ARM architecture 4 cores, max 2.0GHz |
| GPU | Mali G510 MC4 |
| NPU | 1Tops(INT8) |
| RAM | LPDDR4/4X particles 4G |
| Memória | 1 X EMMC 32G, built-in expandable TF card expansion according to demand at the factory |
| Video | 1x HDMI |
| Audio output | 1x ∮3.5mm |
| Rede | 1x RJ45 10/100/1000Mbps |
| Wi-Fi | 1 x board-mounted WIFI, 1 x 4G (configured on demand) |
| USB | 2x USB2.0；1 x USB3.0 |
| Other extension interfaces | 1 x10 pin expansion IO phoenix terminals |
| Power Input | DC 12V 2A |
| Tamanho | 169mm (wide) * 94mm (deep) * 42mm (high) |
| System | Ubuntu 22.04 LTS*4 |
| Work environment | 0~50℃,10%~90% no condensation |
| Storage environment | -20~70℃,5%~90% no condensation |
| Access capability | This device can be connected to 1 channel 4 million or 2 channels 1080p or 3 channels 720p or 4 channels D1 pixel video real-time AI analysis |
| Model No | description |
| SP-5001 | 1Tops/4GB LPDDR4/32GB EMMC/2*USB/1*HDMI, 1*400w pixel video access real-time analysis |
| SP-5002 | 1Tops/4GB LPDDR4/32GB EMMC/2*USB/1*HDMI, 2*1080p pixel video access real-time analysis |
| SP-5003 | 1Tops/4GB LPDDR4/32GB EMMC/2*USB/1*HDMI, 3*720p pixel video access real-time analysis |
| SP-5004 | 1Tops/4GB LPDDR4/32GB EMMC/2*USB/1*HDMI, 4*D1 pixel video access real-time analysis |
| Application scenario | ||
| No | Item | Description |
| 1 | Object category recognition capability | People, fallen individuals, smokers, glasses, backpacks, seat belts, reflective clothing, hats, chef’s caps, safety helmets, mobile phones, masks, handbags, road potholes, masked faces, frontal facial views, side profile shots, raised hands, knives, firearms, sticks, cars, bicycles, motorcycles, vans, buses, minibuses, trains, taxi signs, excavators, dump trucks, dump trucks, container trucks, construction waste trucks, airplanes, kites, dogs, cats, cattle, horses, birds, mice, cockroaches, geckos, suitcases, packing boxes, hands, tricycles, wheelchairs, flames, smoke columns, snakes, and over 200 other common object categories |
| 2 | Mix wire detection | This calculation rule draws a line in the video to specify the category of the surveillance object. When the surveillance object passes the line, the system detects and triggers a capture event. If the object stays on the line for a long time, a capture event is triggered every N seconds, and the user can specify which object or multiple categories to detect
This counting rule is usually used to detect the passage of an object, such as people crossing boundaries, vehicles entering and leaving, animals entering and leaving, etc. |
| 3. | AB line detection | This calculation rule draws two lines in the video to specify the category of the surveillance object. When the surveillance object passes through line A and then line B, the system detects the capture event; otherwise, it does not trigger the event
This rule of the law of motion is usually applied to an object that requires a certain direction of movement, such as people crossing boundaries, vehicles entering and leaving, animals entering and leaving, going against the direction, not driving in the direction, etc.
|
| 4 | Object retention detection | This calculation rule draws a rectangular or polygonal control area in the video, and sets a certain object to stay in the control area for a long time (set more than the specified N seconds), and then generates a capture event
These rules are usually applied to perimeter warning, vehicle parking violations, and personnel wandering |
| 5 | Regional intrusion detection | This calculation rule draws a rectangular or polygonal control area in the video, and specifies that a certain object passes through the control area, tracks and generates a capture event
These accounting rules are usually applied to perimeter alerts |
| 6 | Object removed from detection | This calculation rule draws a rectangular or polygonal control area in the video, and sets that an object must exist in the control area. If the object is not detected within the set time range, a capture event will be generated
These accounting rules are usually applied to the security of off-duty personnel or guards, and the care of valuables |
| 7 | Enter the AB area for detection | This calculation rule draws two areas in the video and specifies the category of the surveillance object. The system detects when the surveillance object passes through area A and then through area B, and triggers the capture event; otherwise, it does not trigger the capture event
This traffic law rule is usually applied to the movement of an object required by the direction, such as people crossing the boundary, vehicles illegally changing lanes, not driving in the direction, going against the direction, etc. |
| 8 | Crowd gathering detection | This calculation rule draws the area in the video, and when the number of people in the area exceeds N, a crowd gathering snap event is generated |
| 9 | Vehicle congestion detection | This calculation rule draws the area in the video and sets the number of vehicles in the area to exceed N, and then generates the vehicle congestion capture event |
| 10 | Regional flow statistics | This accounting rule draws the area in the video and tracks the people entering the area. When the people leave the area, a capture counting event is generated
These rules of the accounting are applied to regional population statistics |
| 11 | Someone fell | This calculation rule draws the area in the video and tracks the people entering the area. When the person falls in the area, a capture event is generated
These rules are usually applied to nursing homes, communities and caregivers to warn of falls. |
| 12 | Object stationary detection [Sleeping guard at the gate] | This calculation rule draws the area in the video and tracks the people entering the area. When the people are stationary, a capture event is generated. This accounting rule is usually used to identify whether a guard/keeper is asleep or not |
| 13 | Fire detection | This calculation rule draws the area in the video and generates a capture event when a flame or smoke column is generated in the controlled area
These fire codes are usually applied to fire safety places |
| 14 | phone | This calculation rule draws the area in the video and identifies the capture event when people make calls in the controlled area
This calculation rule is usually used in the petrochemical industry to prevent the wireless current from triggering a fire when making a phone call; and to prevent some production safety areas from causing safety accidents due to employees making phone calls |
| 15 | No safety helmet identification | This calculation rule draws the area in the video and identifies the capture event when the personnel in the control area does not wear a safety helmet
These rules are usually applied to construction sites and safety workshops to prevent employees from wearing safety helmets in case of accidents |
| 16 | Object rapid mobile detection | This calculation rule draws the area in the video and identifies the capture event when the personnel in the controlled area moves quickly
This accounting rule is usually applied in prisons to give early warning when someone moves quickly |
| 17 | Blacklist face recognition warning | This calculation rule draws the area in the video and identifies whether the personnel in the surveillance area are blacklisted. If the personnel are blacklisted, a capture event will be generated and a timely warning will be issued to notify security personnel to deal with it
This accounting rule is usually applied to units or communities, or public security control, to detect blacklisted personnel in time for staff to deal with |
| 18 | Pedestrians are structured | This algorithm rules draw areas in the video and track and identify whether people entering the area are wearing safety helmets, riding electric two-wheelers, riding bicycles, wearing hats, holding mobile phones, smoking, facing forward, riding tricycles, male/female (face recognition), gender (face recognition), carrying suitcases, or holding luggage |
| 19 | Count the number of people in and out | This counting rule is to draw two lines in the video, namely A line and B line. When people move from A line to B line, a counting event of human flow is triggered. It is usually used in places where directional human flow counting statistics are available |
| 20 | No glove recognition | This calculation rule draws the area in the video and identifies the capture event when the personnel in the controlled area does not wear gloves
This rule of thumb is usually applied to situations where gloves are required to prevent employees from not wearing gloves, which may affect hygiene or other safety incidents |
| 21 | Smoking detection | This calculation rule draws the area in the video and identifies the capture event when a person is smoking in the controlled area
These fire codes are usually applied to fire safety places |
| 22 | Someone with a knife identified | This calculation rule draws the area in the video, and when people are caught holding knives in the controlled area, it will generate a capture event and give a timely warning
This accounting rule is usually applied to campus security monitoring |
| 23 | No mask recognition | This calculation rule draws the area in the video and identifies the capture event when the person in the controlled area does not wear a mask
These rules are usually applied in the kitchen or other work situations where masks are required to prevent employees from not wearing masks, which may affect kitchen hygiene or other safety incidents |
| 24 | No chef’s hat identification | The algorithm rules draw the area in the video and identify the snap event when the personnel in the controlled area does not wear a chef’s hat
These rules are usually applied in the kitchen to prevent employees from not wearing a chef’s hat, which may affect the sanitary environment of the kitchen |
| 25 | Face recognition of foreign personnel | The algorithm rule draws the area in the video and generates a capture event when it is recognized that the personnel in the controlled area is not an internal person
This rule of law is usually applied to units or communities to prevent the entry of idle people |
| 26 | White list face recognition | The algorithm rule draws the area in the video and identifies whether the personnel in the controlled area are internal personnel. If the personnel are internal personnel, the entrance and exit will be allowed to pass, the door will be opened, the attendance record will be recorded, and the meeting will be signed in
This rule of law is usually applied to units or communities to prevent the entry of idle people |
| 27 | Fighting behavior identification | This calculation rule supports the identification of fighting behavior and timely warning |
| 28 | Not identified by reflective clothing | Whether the inspector is wearing a reflective clothing or whether any member of the party is wearing reflective clothing |
| 29 | No seat belt identification | Whether the inspector is not wearing a seat belt |
| 30 | Petroleum and stone chemical manual service identification | Whether the testing personnel wear work clothes generally needs custom development, because the work clothes of each unit may be different |
| 31 | Ride without a helmet | Support testing of people who ride bicycles, electric cars and tricycles without helmets |
| 32 | Hold up your hand for help | Support the way the test personnel raise their hands to send out a distress signal |
| 33 | The van was illegally carrying passengers | Support the detection of illegal passenger carrying of all kinds of trucks, or the capture of passenger carrying on household tricycles |
| 34 | Car license plate recognition | Supports various vehicle license plate recognition, compatible with new energy vehicle license plate recognition, compatible with Hong Kong license plate recognition |
| 35 | License plate recognition for two wheel electric vehicles | Support various automatic motorcycle/cycle white plate, yellow plate, blue plate identification |
| 36 | Limited number of participants | Supports setting a warning for the number of people less than or greater than the specified number in the video scene |
| 37 | Parking statistics | Supports setting the number of parking Spaces in the parking area and how many parking Spaces have been used |
| 38 | Ride with a person | Support the detection of people carrying on two-wheeled motorcycles and electric motorcycles, and automatically filter out events of cycling |
| 39 | Not wearing protective clothing | Support the testing of special protective clothing for epidemic prevention, set up testing areas, and take pictures to warn when people are found not wearing protective clothing |
| 40 | Traffic accident-car crash | Support vehicle collision events on the road during detection and generate capture and warning events |
| 41 | Traffic accident-vehicle rollover | Support vehicle rollover events on the road during detection and generate capture warning events |
| 42 | Traffic accident-electric car rollover | Support the detection of motorcycle or electric bicycle rollover events on the road, and generate snap alert events |
| 43 | Riding a tricycle without a helmet | Support the detection of unhelmeted tricycle riders on the road and generate snap alert events |
| 44 | Interval speed measurement | In the single video scene detection, set A section and B section to detect the speed (m/s or km/h) of vehicles or other objects from A to B, and support setting of overspeed warning |
| 45 | Electric cars run red lights | The system can recognize the status of traffic lights, capture and identify the behavior of electric two-wheelers running red lights when the lights are red, support the identification of electric vehicle license plates, and link to video evidence collection |
| 46 | Video blocking | When the video is black screen or covered, an alert is generated |
| 47 | Garbage sorting inspection | The system automatically identifies garbage piles, garbage bags, boxes, bottles, packing cases, packaging boxes, etc |
| 48 | Road disease detection | The system supports the detection of various road surface diseases, such as cracks and potholes |
| 49 | Coal mine conveyor belt detection | Support the identification of belt coal, no coal, more coal and less coal states, belt offset, state identification, identification of foreign objects such as anchor rods, gangue, wooden boards, woven bags on the belt |
| 50 | Coal mine conveyor belt detection | Support the identification of belt coal, no coal, more coal and less coal states, belt offset, state identification, identification of foreign objects on the belt such as anchor rods, gangue, wooden boards, woven bags, etc |
| 51 | Large vehicles are photographed without stopping at the intersection | Automatically capture the behavior of trucks turning right at intersections without stopping to observe, and identify the license plate number |
| 52 | Carry trash by hand | Automatic recognition and identification of carrying out garbage is usually used for voice prompt or linkage to open the lid of the trash can before throwing away garbage |
| 53 | Electromagnetic coil test (made by Midea) | Used in the production test process, abnormal products can not be put into the qualified product storage box |
| 54 | Abnormal oil discharge detection at gas station | Real-time analysis and identification of bad behaviors of personnel in the gas station (such as smoking, talking on the phone, spontaneous combustion, etc.) and whether the staff is off duty during oil unloading, whether the static clip is connected, whether the fire extinguisher is placed, whether the oil guide pipe is connected and other actions are automatically identified, and alarm is given when abnormal |
| 55 | Play with your phone | Automatically identify and capture employees’ behavior of playing with their phones during working hours |
| 56 | The sludge scraping arm of the sewage plant’s sinking tank is stopped | The automatic detection of the sludge scraping arm movement in the sinking tank of the sewage plant will generate an early warning event after N seconds |
| 57 | Coal mine conveyor belt displacement | The automatic identification transmission belt shift N seconds later generates an early warning event |
| 58 | Coal mine conveyor belt flow | Automatically identify the flow rate of coal transported by conveyor belt and output the result in percentage |
| 59 | Coal mine conveyor belt blockage | The automatic identification transmission belt blockage generates an early warning event |
| 60 | Customized screening of prison cell personnel | The personnel in the room counted and tested according to the type of clothing |
| 61 | Test for garbage not in bag | Garbage classification and disposal points, kitchen waste mixed input detection |
| 62 | Vertical intersection operation warning | Vertical operation detection of upper and lower struts on the construction site |
| 63 | Abnormal iron chain in coal mine comprehensive mining [custom] | Abnormal iron chain detection of automatic coal mining machine in coal mine |
| 64 | Highway parking inspection | Highway detection of a car suddenly stopped warning capture |
| 65 | Fire lanes blocked | Fire protection usually occupies the detection |
| 66 | The bin is overflowing | Garbage classification and disposal point, garbage can overflow detection |
| 67 | Highway toll station passage pedestrian safety | The pedestrian safety detection of the expressway toll station is linked, and the pole is locked when passing, and unlocked after the person passes |
| 68 | No wristbands | At the factory, employees on the operating table are not tested with electrostatic wristbands |
| 69 | Highway pedestrian incidents | Highway pedestrian detection and capture warning |
| 70 | Highway non-motor vehicle incidents | Non-motor vehicle detection and capture warning on expressways |
| 71 | Not wearing a life ring | Check whether the people in the mine are wearing life-saving bells |
| 72 | The baffle plate is abnormal | Check whether the support plate of the underground fully mining machine in the coal mine is fallen or damaged |
| 73 | Handmade job operation time efficiency detection | Monitor the operation rate of employees in the factory operation desk when working manually |
| 74 | Coal trolley count statistics | Statistics on the number of coal trucks coming out of the well |
| 75 | Object traffic statistics | Specifies the object type and counts the traffic of passing objects in the scenario |
| 76 | Object quantity statistics | Specifies the object type and counts the number of objects in the scene |
| 77 | Object feature comparison | Detects whether the specified object exists in the scene or does not exist in the scene |
| 78 | Multi-scenario object quantity detection | Detect the total number of specified object categories in multiple camera video scenes |
| 79 | Don’t leash your dog | Detect the event of capturing a dog running without a leash in the video |
| 80 | Pull the banner | When someone pulls a banner in the detection video, a snapshot event is generated |
| 81 | The vehicle did not yield to the pedestrian at the crosswalk | The detection video shows that there is a capture event when a vehicle does not yield to pedestrians on the crosswalk |
| 82 | Jump over a barrier or fence | Detect the capture event when someone climbs over the fence or pedestrian gate in the video |
| 83 | Approach or leave an object | When an object in the area set in the detection video approaches or leaves the object, an early warning event is generated |
| 84 | No hat | When the person in the set area of the detection video does not wear a hat, an early warning event is generated |
| 85 | Custom model object detection | This rule is used to load a small model for specific training and generate objects when detecting specific scene objects. It is usually used to extend detection of some scenarios with relatively simple business functions. |
| 86 | Abnormal behavior in class or during a lesson | When people in the set area of the detection video whisper, stand, reach out to pass, turn their heads and leave their seats, an early warning event is generated |
| 87 | The vandalism | When the person in the set area of the detection video holds a weapon and breaks things up, an early warning event is generated |
| 88 | Fire door not closed | When the fire door in the detection video is not closed for N seconds, a capture event is generated |
| 89 | Safety production sequence identification | In the process of production safety integration, various behaviors are set, trigger logic conditions are set, abnormal behaviors in the video are detected, and early warning events are generated |
| 90 | Supervision and scoring of kitchen waste disposal | There are two modes to detect the quality of kitchen waste in community garbage disposal stations: 1. Manual scoring (three buttons “Excellent”, “Good” and “Poor” are obtained by the analysis equipment to capture events) and 2. Automatic analysis scoring (the AI system automatically detects the quality of the thrown garbage and generates three levels of captured events “Excellent”, “Good” and “Poor”) |
| 91 | No shoe covers | When the person in the set area of the detection video does not wear a shoe cover, an early warning event is generated |
| 92 | There was a drowning incident | When someone is drowning in the area set in the detection video, an early warning event occurs |
| 93 | Garbage disposal station is tested regularly | For the garbage disposal station, the scheduled detection is set up to generate event types such as normal, exposed garbage, overflow, displacement, poor image quality, garbage container not closed, trash can outside the swing, and vehicle occupying the disposal area. |
| 94 | fatigue detection | When the close-range personnel in the detection video have fatigue behaviors such as prolonged closed eyes and yawning, an early warning event will be generated |
| 95 | Not in the designated area | When a pedestrian does not enter the set area to perform an action (such as washing hands, touching the static ball, etc.), an early warning event is generated |
| 96 | No life jacket worn | Detect the warning event when the pedestrian does not penetrate the life jacket |
| 97 | The stationary object begins to move | When a stationary object under detection starts to move, an alert event is generated |
| 98 | Coal leakage from coal mine conveyor belt | When coal leakage occurs on the conveyor belt of the mine, an early warning event is generated |
| 99 | Coal conveyor belt water coal | Early warning events are generated when water and coal are detected on coal conveyor belts in mines |
Português 
English 
Deutsch 
Русский 
Français 
Español 
Italiano 
简体中文 
हिन्दी 
العربية 
বাংলা 
Bahasa Indonesia 
日本語 
Tiếng Việt 
Türkçe 
ไทย 
فارسی 
Bahasa Melayu