Animal Identification Intelligent Camera (AIIC)
Biological resources are an important factor to maintain the balance of ecosystem, and animal resources occupy an important position in biological resources. Protecting animal resources plays an important role in maintaining ecosystems. Nowadays, endangered wild animals need regular protection from loss of biodiversity. In order to better protect animal resources, intelligent camera can detect the animals in real time and classify and locate the animals in the image.
Product Feature:
1.This product can be used in multiple environments, Users are free to train and deploy specific models to deal with animals in different environments.
2.The intelligent camera is calculated in the camera, there is no need to send back the image to the back end for calculation. So a backend device can be used to receive data from multiple smart cameras.
3.The intelligent camera is equipped with 4G modules, which can transmit data wirelessly and control it remotely.
4.Intelligent cameras don't need real-time attention from users. When an animal is detected, it will automatically save the video from one minute before the animal was detected, and from one minute after the animal left.
5.Intelligent camera has functions such as real-time monitoring, reviewing animal videos and recording the time and place of animal appearance. Also can operate on multiple cameras in the same backend system.
Key Technology :
The size of the animal target in the image is affected by the near and far angle and so on. Animals of the same species may vary in size, and may be small near or far away, resulting in the small size of common large animals in the picture. In the process of detection, the detection accuracy of detection algorithm will be reduced due to the influence of different animal target sizes. Therefore, by fusing the multi-scale convolution feature and introducing the multi-scale context information, we solve the problem of false detection and missing detection caused by the phenomenon of too small size of animal targets in the image. The method of feature fusion can enhance the semantic information of low-level features and enhance the detailed information of high-level features. At the same time, multi-scale context information processing is introduced to indirectly help identify the animal target information in the diagram, thus improving the detection accuracy of animal targets.
Camera hardware configuration:
2 million star level 1 / 1.8 "CMOS D-SHORE AI open platform barrel network camera;
High efficiency white light array lamp is built in the device, low power consumption;
The device has built-in electric zoom lens, easy to operate and stable zoom process;
Support the distribution and operation of AI model, generate test results and upload to the business platform;
It supports the detection of specific targets and classifies the detection results, which can be uploaded to the business platform;
Support the video task, analyze the real-time video, analyze according to the set frame rate, and upload the results according to the set alarm interval;
Support capture round patrol task, according to the set time interval capture analysis, and upload the results according to the set alarm interval;
It supports 4 model package storage, and each model package supports 1 Detection Model and 4 classification models;
It supports 16 kinds of target detection and classifies 4 kinds of targets;
It supports the model scheduling function in the capture task mode, and supports the serial execution of up to 4 model packages;
It supports the configuration of event rules, and the detection and classification results are filtered according to the rules set to generate alarms;
Support area detection target statistics and cross-line statistics, statistical results superimpose OSD;
Minimum illumination: Color: 0.0005lux @ (f1.2, AGC on) black and white: 0.0001lux @ (f1.2, AGC on), 0 lux with IR;
Lens: (zoom) 8-32mm @ f1.8, horizontal FOV: 37.7 ° ~ 15.1 °, vertical FOV: 21.0 ° ~ 8.7 °, diagonal FOV: 43.4 ° ~ 17.3 °;
Wide dynamic: ultra wide dynamic range up to 120dB, indoor backlight environment monitoring;Video compression standard: h.265/h.264/mjpeg;
Maximum image size: 1920 × 1080;
Storage function: support micro SD (i.e. TF Card) / micro SDHC / micro SDXC card (128GB or 256gb) disconnected local storage and continuous transmission, NAS (NFS, SMB / CIFS support), support SD card encryption and SD status detection function with D-SHORE black card;
Interface type: swing line;
Communication interface: one RJ45 10m / 100M / 1000m adaptive Ethernet port; RS-485 (half duplex, D-SHORE, pelco-p, PELCO-D, self adaptive);
Power output: DC12V 200mA;
Audio interface: audio input: support 2-way 3.5mm jack line in; audio output: support 1-way 3.5mm jack line out;
Alarm interface: 3 inputs, 2 outputs (alarm output maximum support AC / DC24V 1a);
Reset button: support;
Working temperature and humidity: - 30 ºC ~ 60 ºC, humidity less than 95% (no condensation);Power supply: DC: 12V ± 20%, support anti reverse connection protection; POE: 802.3at, class 4;
Power interface type: three core power interface;
Power consumption: dc:12v, 1.3A, max:15.0w; poe: (802.3at, 42.5v-57v), 0.3A to 0.4A, max:17.0w;
Protection grade: IP66;
Distance of supplementary light: white light: 80m;
Product size (mm): 206.5 × 103.9 × 100 mm;
Package size (mm): 385 × 158 × 155mm;
Bare weight: fuselage weight: 1450g;
Packing weight: with packing weight: 2300G;
Biological resources are an important factor to maintain the balance of ecosystem, and animal resources occupy an important position in biological resources. Protecting animal resources plays an important role in maintaining ecosystems. Nowadays, endangered wild animals need regular protection from loss of biodiversity. In order to better protect animal resources, intelligent camera can detect the animals in real time and classify and locate the animals in the image.
Product Feature:
1.This product can be used in multiple environments, Users are free to train and deploy specific models to deal with animals in different environments.
2.The intelligent camera is calculated in the camera, there is no need to send back the image to the back end for calculation. So a backend device can be used to receive data from multiple smart cameras.
3.The intelligent camera is equipped with 4G modules, which can transmit data wirelessly and control it remotely.
4.Intelligent cameras don't need real-time attention from users. When an animal is detected, it will automatically save the video from one minute before the animal was detected, and from one minute after the animal left.
5.Intelligent camera has functions such as real-time monitoring, reviewing animal videos and recording the time and place of animal appearance. Also can operate on multiple cameras in the same backend system.
Key Technology :
The size of the animal target in the image is affected by the near and far angle and so on. Animals of the same species may vary in size, and may be small near or far away, resulting in the small size of common large animals in the picture. In the process of detection, the detection accuracy of detection algorithm will be reduced due to the influence of different animal target sizes. Therefore, by fusing the multi-scale convolution feature and introducing the multi-scale context information, we solve the problem of false detection and missing detection caused by the phenomenon of too small size of animal targets in the image. The method of feature fusion can enhance the semantic information of low-level features and enhance the detailed information of high-level features. At the same time, multi-scale context information processing is introduced to indirectly help identify the animal target information in the diagram, thus improving the detection accuracy of animal targets.
Camera hardware configuration:
2 million star level 1 / 1.8 "CMOS D-SHORE AI open platform barrel network camera;
High efficiency white light array lamp is built in the device, low power consumption;
The device has built-in electric zoom lens, easy to operate and stable zoom process;
Support the distribution and operation of AI model, generate test results and upload to the business platform;
It supports the detection of specific targets and classifies the detection results, which can be uploaded to the business platform;
Support the video task, analyze the real-time video, analyze according to the set frame rate, and upload the results according to the set alarm interval;
Support capture round patrol task, according to the set time interval capture analysis, and upload the results according to the set alarm interval;
It supports 4 model package storage, and each model package supports 1 Detection Model and 4 classification models;
It supports 16 kinds of target detection and classifies 4 kinds of targets;
It supports the model scheduling function in the capture task mode, and supports the serial execution of up to 4 model packages;
It supports the configuration of event rules, and the detection and classification results are filtered according to the rules set to generate alarms;
Support area detection target statistics and cross-line statistics, statistical results superimpose OSD;
Minimum illumination: Color: 0.0005lux @ (f1.2, AGC on) black and white: 0.0001lux @ (f1.2, AGC on), 0 lux with IR;
Lens: (zoom) 8-32mm @ f1.8, horizontal FOV: 37.7 ° ~ 15.1 °, vertical FOV: 21.0 ° ~ 8.7 °, diagonal FOV: 43.4 ° ~ 17.3 °;
Wide dynamic: ultra wide dynamic range up to 120dB, indoor backlight environment monitoring;Video compression standard: h.265/h.264/mjpeg;
Maximum image size: 1920 × 1080;
Storage function: support micro SD (i.e. TF Card) / micro SDHC / micro SDXC card (128GB or 256gb) disconnected local storage and continuous transmission, NAS (NFS, SMB / CIFS support), support SD card encryption and SD status detection function with D-SHORE black card;
Interface type: swing line;
Communication interface: one RJ45 10m / 100M / 1000m adaptive Ethernet port; RS-485 (half duplex, D-SHORE, pelco-p, PELCO-D, self adaptive);
Power output: DC12V 200mA;
Audio interface: audio input: support 2-way 3.5mm jack line in; audio output: support 1-way 3.5mm jack line out;
Alarm interface: 3 inputs, 2 outputs (alarm output maximum support AC / DC24V 1a);
Reset button: support;
Working temperature and humidity: - 30 ºC ~ 60 ºC, humidity less than 95% (no condensation);Power supply: DC: 12V ± 20%, support anti reverse connection protection; POE: 802.3at, class 4;
Power interface type: three core power interface;
Power consumption: dc:12v, 1.3A, max:15.0w; poe: (802.3at, 42.5v-57v), 0.3A to 0.4A, max:17.0w;
Protection grade: IP66;
Distance of supplementary light: white light: 80m;
Product size (mm): 206.5 × 103.9 × 100 mm;
Package size (mm): 385 × 158 × 155mm;
Bare weight: fuselage weight: 1450g;
Packing weight: with packing weight: 2300G;