{"data":{"id":"145686","string_id":"MT62dV","status":"1","parent_id":"0","slug":"MT62dV","order":"5","name":"User Guide","type":"1","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"","seo_keyword":"","seo_desc":"","author":"","published_time":"0","add_time":"1724144130","update_time":"1749802414","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"1","tags":[],"child":[{"id":"178778","string_id":"mUISanIc","status":"1","parent_id":"145686","slug":"intf","order":"7","name":"\ud83d\udd17Connecter Interface","type":"2","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"Connecter Interface","seo_keyword":"","seo_desc":"","share_switch":"0","share_code":"","author":"","published_time":"1728616721","add_time":"1728616615","update_time":"1759211120","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"3","file_type":"","auth_id":"[\"-1\"]","content":{"content_id":"178778","content":"

$\"\"$ <\/h2>\n

1. PTZ Camera Interface<\/h2>\n
$\"\"$ \n
\n
Aviation plug interface<\/p>\n<\/figcaption>\n<\/figure>\n
<\/p>\n
$\"\"$ \n
\n
Cable Definition<\/p>\n<\/figcaption>\n<\/figure>\n
The plug interface definition is shown in the table.<\/p>\n
\n\n\n\n\n\n\n\n\n\n\n\n\n
No.<\/th>\n Define<\/th>\n<\/tr>\n<\/thead>\n
1<\/td>\n Power In<\/td>\n<\/tr>\n
2<\/td>\n GND<\/td>\n<\/tr>\n
3<\/td>\n RS485-<\/td>\n<\/tr>\n
4<\/td>\n CVBS Video<\/td>\n<\/tr>\n
5<\/td>\n RS485+<\/td>\n<\/tr>\n
6<\/td>\n ETH TD+<\/td>\n<\/tr>\n
7<\/td>\n ETH TD-<\/td>\n<\/tr>\n
8<\/td>\n ETH RD+<\/td>\n<\/tr>\n
9<\/td>\n ETH RD-<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
2. AI Module Interface<\/h2>\n
Interface definition<\/h3>\n
<\/p>\n
<\/colgroup>\n\n\n\n\n\n
Connecter No.<\/strong><\/td>\n Connection<\/strong><\/td>\n Description<\/strong><\/td>\n<\/tr>\n
1<\/td>\n Power In\/out<\/td>\n DC12V<\/td>\n<\/tr>\n
2<\/td>\n Ethernet<\/td>\n 100 based ETH<\/td>\n<\/tr>\n
3<\/td>\n RS485<\/td>\n Can connect to Pan Tilt or MODBUS device<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Ethernet connector<\/h3>\n
We provide waterproof kits, connect the network cable as shown in the figure.<\/p>\n
<\/p>\n
<\/p>\n
<\/p>\n
Power In\/Out Connector<\/h3>\n
The specification of the power connector is 5525, and the connection method is shown in the following figure.<\/p>\n
<\/p>\n
<\/p>\n
RS485 Connector<\/h3>\n
If there is a gimbal that needs to be controlled, it can be connected to the RS485 interface.<\/p>\n
<\/p>\n
<\/colgroup>\n\n\n\n\n
Label<\/strong><\/td>\n Define<\/strong><\/td>\n Description<\/strong><\/td>\n<\/tr>\n
+<\/span><\/strong><\/td>\n RS485 A<\/td>\n 9600 bps baud rate<\/td>\n<\/tr>\n
-<\/span><\/strong><\/td>\n RS485 B<\/td>\n <\/td>\n<\/tr>\n<\/tbody>\n<\/table>","pure_content":"1. PTZ Camera Interface\n\n\nAviation plug interface\n\n\n\u00a0\n\n\nCable Definition\n\n\nThe plug interface definition is shown in the table.\n\n\n\n No.\n Define\n\n\n\n\n 1\n Power In\n\n\n 2\n GND\n\n\n 3\n RS485-\n\n\n 4\n CVBS Video\n\n\n 5\n RS485+\n\n\n 6\n ETH TD+\n\n\n 7\n ETH TD-\n\n\n 8\n ETH RD+\n\n\n 9\n ETH RD-\n\n\n\n2. AI Module Interface\nInterface definition\n\n\n\n\n Connecter No.\n Connection\n Description\n\n\n 1\n Power In\/out\n DC12V\n\n\n 2\n Ethernet\n 100 based ETH\n\n\n 3\n RS485\n Can connect to Pan Tilt or MODBUS device\n\n\n\nEthernet connector\nWe provide waterproof kits, connect the network cable as shown in the figure.\n\n\n\u00a0\nPower In\/Out Connector\nThe specification of the power connector is 5525, and the connection method is shown in the following figure.\n\n\nRS485 Connector\nIf there is a gimbal that needs to be controlled, it can be connected to the RS485 interface.\n\n\n\n\n Label\n Define\n Description\n\n\n +\n RS485 A\n 9600 bps baud rate\n\n\n -\n RS485 B\n \u00a0","is_draft":"0","editor_type":"3","file_path":"","add_time":"1728616615","update_time":"1728637230"},"pure_content":"1. PTZ Camera Interface\n\n\nAviation plug interface\n\n\n\u00a0\n\n\nCable Definition\n\n\nThe plug interface definition is shown in the table.\n\n\n\n No.\n Define\n\n\n\n\n 1\n Power In\n\n\n 2\n GND\n\n\n 3\n RS485-\n\n\n 4\n CVBS Vide"},{"id":"178792","string_id":"kxfxQB8A","status":"1","parent_id":"145686","slug":"finetune","order":"6","name":"\u2699\ufe0fTracking and PTZ Parameter Fine-tune Guide","type":"2","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"Tracking and PTZ Parameter Fine-tune Guide","seo_keyword":"","seo_desc":"","share_switch":"0","share_code":"","author":"","published_time":"1728618797","add_time":"1728618268","update_time":"1767750717","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"2","file_type":"","auth_id":"[\"-1\"]","content":{"content_id":"178792","content":"
$\"\"$
<\/figcaption><\/figure><\/center>\n\n>The fine-tuning process is only supported under the LPP\/ExPelcoD protocol.\n>\n\n![](https:\/\/resource-wangsu.helplook.net\/docker_production\/4lnqex\/article\/kxfxQB8A\/68d50b2943110.png)\n\n\nAt this point, the custom parameter area is used.\nThere are a total of 8 custom parameters, and only param4, 6, and 7 parameters are used in the fine-tuning process. Please do not modify any other parameters.\n\nThe steps are as follows:\n- 1. First, ensure that the Zoom list tableof the device camera is matched and imported into the Tofu device.\n- 2. Use the ROI tracking mode, set the Max speed to 65535 under static background, that is, unlimited state. Set the Diff Coef to 1000, and do not set the Speed gradient and Int Coef to default. Now, enter the fine tuning static parameter mode.\nIn the tracking mode of ROI tracking, adjust the lens magnification to 1x. Par4 is the value obtained by rounding the . The specific value is determined by the gimbal's own speed configuration. Maintain a fixed value in the future.\n- 3. par6 and par7 are in a 1:10 ratio, with par7 as an interval of 100, gradually rising. Track the objects, in boundary of the frame area and 1\/4 of the frame, to see the tracking effect. After setting the parameters, you need to click \"Update\" to take effect. After obtaining the approximate effect, fine adjustments can be made at intervals of 10 or 100.\n- 4. When the selected object in the screen can be centered within 2 seconds without losing tracking, it means that the par7 parameter has been adjusted.\n- 5. Adjust par6 within the range of 1\/5 to 1\/20. If there is oscillation after centering, reduce it; if the centering is not sufficient, increase it.\n- 6. Taking the gimbal subdivided by 0.01\u00b0 as an example, the empirical values of par6 and par7 are 5 and 260 respectively;\nTaking the 0.001\u00b0 subdivision of the gimbal as an example, the empirical values of par6 and par7 are 400 and 3600. The above data is for reference only.\n- 7. After adjusting par6 and par7 in the above steps, update and save the parameters Adjust the Diff Coef to 0, which means turn off the static parameters. Start adjusting dynamic parameters.\nIt is recommended to adjust dynamic parameters when there is a moving target, and parameters tested in static scenes can be fine tuned under the moving target to better adapt.\n- 8. Speed gradient increases gradually at intervals of 100, and tracks objects, in the boundary of the frame and 1\/4 of the frame. Check the object's centering time and overshoot status. Overshoot refers to the phenomenon where the selected target continues to move away from the center point after reaching the center position. When there is a slight overshoot, the increase in motion coefficient can be stopped. When the speed subdivision is not small enough, the motion coefficient can be reduced and the interval increased.\nThe motion coefficient is generally similar to par7.\n- 9. Increase the Int Coef to adjust the centering of the target, which can generally be set to 1\/10 or 1\/20 of the Speed gradient.\n- 10. The maximum speed is limited to 3-10 times the Speed gradient.\n- 11. Change the Diff Coef to 50.\n\n>If you want to achieve excellent tracking centering and use a third-party servo, we can provide you with independent servo precise parameter adjustment technical services. Accurate parameter tuning technology services require remote operation after our technical evaluation or the customer providing physical equipment, replacing independent precision tuning tracking algorithms, and providing them after parameter calibration. This service is a one-time service, and there is no need to recalibrate the same type of servo\n\n### Online Auto Param Culc Software: \n
$\"\"$
<\/figcaption><\/figure>\n\n[Click Here to Tofu LensCulc](https:\/\/tofulens.streamlit.app\/)\n
\n\n---\n
$\"\"$
<\/figcaption><\/figure><\/center>","pure_content":"The fine-tuning process is only supported under the LPP\/ExPelcoD protocol.\n\n\nAt this point, the custom parameter area is used.\nThere are a total of 8 custom parameters, and only param4, 6, and 7 parameters are used in the fine-tuning process. Please do not modify any other parameters.\nThe steps are as follows:\n\n\n\nFirst, ensure that the Zoom list tableof the device camera is matched and imported into the Tofu device.\n\n\n\n\nUse the ROI tracking mode, set the Max speed to 65535 under static background, that is, unlimited state. Set the Diff Coef to 1000, and do not set the Speed gradient and Int Coef to default. Now, enter the fine tuning static parameter mode.\nIn the tracking mode of ROI tracking, adjust the lens magnification to 1x. Par4 is the value obtained by rounding the . The specific value is determined by the gimbal's own speed configuration. Maintain a fixed value in the future.\n\n\n\n\n\npar6 and par7 are in a 1:10 ratio, with par7 as an interval of 100, gradually rising. Track the objects, in boundary of the frame area and 1\/4 of the frame, to see the tracking effect. After setting the parameters, you need to click \"Update\" to take effect. After obtaining the approximate effect, fine adjustments can be made at intervals of 10 or 100.\n\n\n\n\nWhen the selected object in the screen can be centered within 2 seconds without losing tracking, it means that the par7 parameter has been adjusted.\n\n\n\n\nAdjust par6 within the range of 1\/5 to 1\/20. If there is oscillation after centering, reduce it; if the centering is not sufficient, increase it.\n\n\n\n\nTaking the gimbal subdivided by 0.01\u00b0 as an example, the empirical values of par6 and par7 are 5 and 260 respectively;\nTaking the 0.001\u00b0 subdivision of the gimbal as an example, the empirical values of par6 and par7 are 400 and 3600. The above data is for reference only.\n\n\n\n\nAfter adjusting par6 and par7 in the above steps, update and save the parameters Adjust the Diff Coef to 0, which means turn off the static parameters. Start adjusting dynamic parameters.\nIt is recommended to adjust dynamic parameters when there is a moving target, and parameters tested in static scenes can be fine tuned under the moving target to better adapt.\n\n\n\n\n\nSpeed gradient increases gradually at intervals of 100, and tracks objects, in the boundary of the frame and 1\/4 of the frame. Check the object's centering time and overshoot status. Overshoot refers to the phenomenon where the selected target continues to move away from the center point after reaching the center position. When there is a slight overshoot, the increase in motion coefficient can be stopped. When the speed subdivision is not small enough, the motion coefficient can be reduced and the interval increased.\nThe motion coefficient is generally similar to par7.\n\n\n\n\nIncrease the Int Coef to adjust the centering of the target, which can generally be set to 1\/10 or 1\/20 of the Speed gradient.\n\n\n\n\nThe maximum speed is limited to 3-10 times the Speed gradient.\n\n\n\n\nChange the Diff Coef to 50.\n\n\n\n\nIf you want to achieve excellent tracking centering and use a third-party servo, we can provide you with independent servo precise parameter adjustment technical services. Accurate parameter tuning technology services require remote operation after our technical evaluation or the customer providing physical equipment, replacing independent precision tuning tracking algorithms, and providing them after parameter calibration. This service is a one-time service, and there is no need to recalibrate the same type of servo\n\nOnline Auto Param Culc Software:\n\nClick Here to Tofu LensCulc","is_draft":"0","editor_type":"2","file_path":"","add_time":"1728618268","update_time":"1767750717"},"pure_content":"The fine-tuning process is only supported under the LPP\/ExPelcoD protocol.\n\n\nAt this point, the custom parameter area is used.\nThere are a total of 8 custom parameters, and only param4, 6, and 7 par"},{"id":"527990","string_id":"VqkVMDOP","status":"1","parent_id":"145686","slug":"Custom-Parameters-Setting","order":"5","name":"\ud83d\udddc\ufe0fCustom Parameters Setting","type":"2","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"","seo_keyword":"","seo_desc":"","share_switch":"0","share_code":"","author":"","published_time":"1749802577","add_time":"1749802404","update_time":"1767750667","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"2","file_type":"","auth_id":"[\"-1\"]","content":{"content_id":"527990","content":"
$\"\"$
<\/figcaption><\/figure><\/center>\n\n
$\"\"$
<\/figcaption><\/figure>\n\n### Online Auto Param Culc Software: \n
$\"\"$
<\/figcaption><\/figure>\n\n[Click Here to Tofu LensCulc](https:\/\/tofulens.streamlit.app\/)\n
\n\n\ud83d\udfe2 \u200bCustom Parameter 1 SOT Mode:\u200b\u200b\nSet to 1 for Small Target Tracking Mode, set to 0 for Standard Tracking Mode.\n
\n\n\ud83d\udfe2 \u200bCustom Parameter 2 Min pixel tresh:\u200b\u200b\nMinimum target pixel count for recognition. Targets with pixel counts below this value will be filtered out. Default value is 0, meaning no minimum restriction.\n
\n\n\ud83d\udfe2 \u200bCustom Parameter 3 Max pixel tresh:\u200b\u200b\nMaximum target pixel count for recognition. Targets with pixel counts exceeding this value will be filtered out. Default value is 500, meaning targets above 500 pixels are filtered out.\n
\n\n\ud83d\udfe2 \u200bCustom Parameter 4 Speed\/FOV ratio:\u200b\u200b\n- \u200bSingle Video Input Mode:\u200b\u200b\nParameter 4 represents \u200b10 \u00d7 (Horizontal Subdivision Speed \/ Vertical Subdivision Speed)\u200b. Default is 10 (horizontal and vertical speeds are equal). Use only when speeds differ significantly; avoid arbitrary modification.\n- \u200bDual Video Input Mode:\u200b\u200b\nVisible light magnification factor \u00d7 10 when the fixed-focus thermal infrared and visible light share the same Field of View (FoV). E.g., if thermal FoV matches visible light at 3.2X zoom, input 32.\n
\n\n\ud83d\udfe2 \u200bCustom Parameter 5 Gimbal protocol:\u200b\u200b\n- \u200bFirmware v2.0.9.8:\u200b\u200b\n\u200bparam5\u200b > \u200b0: Continuous Variable Speed Servo Control Mode.\n\u200bparam5\u200b = \u200b0: \u200bPelco-D\u200b Protocol Tracking Mode.\n- \u200bFirmware v2.1.3.9:\u200b\u200b\n\u200bparam5\u200b = \u200b17: Continuous Variable Speed Servo Control Mode.\n\u200bparam5\u200b = \u200b0: \u200bPelco-D\u200b Protocol Tracking Mode.\n
\n\n\ud83d\udfe2 \u200bCustom Parameter 6 & 7 static param:\u200b\u200b\nLPP protocol gimbal setting.Please see [Tracking and PTZ Parameter Fine-tune Guide](https:\/\/tofuai.helplook.net\/docs\/finetune)\n
\n\n\ud83d\udfe2 \u200bCustom Parameter 8 Detect tresh:\u200b\u200b\nRecognition confidence threshold percentage. Default is 20, meaning targets with confidence scores \u2265 20% are identified. Adjust based on specific scenarios.\n\n---\n
$\"\"$
<\/figcaption><\/figure><\/center>","pure_content":"Online Auto Param Culc Software:\n\nClick Here to Tofu LensCulc\n\n\ud83d\udfe2 \u200bCustom Parameter 1 SOT Mode:\u200b\u200b\nSet to 1 for Small Target Tracking Mode, set to 0 for Standard Tracking Mode.\n\n\ud83d\udfe2 \u200bCustom Parameter 2 Min pixel tresh:\u200b\u200b\nMinimum target pixel count for recognition. Targets with pixel counts below this value will be filtered out. Default value is 0, meaning no minimum restriction.\n\n\ud83d\udfe2 \u200bCustom Parameter 3 Max pixel tresh:\u200b\u200b\nMaximum target pixel count for recognition. Targets with pixel counts exceeding this value will be filtered out. Default value is 500, meaning targets above 500 pixels are filtered out.\n\n\ud83d\udfe2 \u200bCustom Parameter 4 Speed\/FOV ratio:\u200b\u200b\n\n\u200bSingle Video Input Mode:\u200b\u200b\nParameter 4 represents \u200b10 \u00d7 (Horizontal Subdivision Speed \/ Vertical Subdivision Speed)\u200b. Default is 10 (horizontal and vertical speeds are equal). Use only when speeds differ significantly; avoid arbitrary modification.\n\u200bDual Video Input Mode:\u200b\u200b\nVisible light magnification factor \u00d7 10 when the fixed-focus thermal infrared and visible light share the same Field of View (FoV). E.g., if thermal FoV matches visible light at 3.2X zoom, input 32.\n\n\n\n\ud83d\udfe2 \u200bCustom Parameter 5 Gimbal protocol:\u200b\u200b\n\n\u200bFirmware v2.0.9.8:\u200b\u200b\n\u200bparam5\u200b > \u200b0: Continuous Variable Speed Servo Control Mode.\n\u200bparam5\u200b = \u200b0: \u200bPelco-D\u200b Protocol Tracking Mode.\n\u200bFirmware v2.1.3.9:\u200b\u200b\n\u200bparam5\u200b = \u200b17: Continuous Variable Speed Servo Control Mode.\n\u200bparam5\u200b = \u200b0: \u200bPelco-D\u200b Protocol Tracking Mode.\n\n\n\n\ud83d\udfe2 \u200bCustom Parameter 6 & 7 static param:\u200b\u200b\nLPP protocol gimbal setting.Please see Tracking and PTZ Parameter Fine-tune Guide\n\n\ud83d\udfe2 \u200bCustom Parameter 8 Detect tresh:\u200b\u200b\nRecognition confidence threshold percentage. Default is 20, meaning targets with confidence scores \u2265 20% are identified. Adjust based on specific scenarios.","is_draft":"0","editor_type":"2","file_path":"","add_time":"1749802404","update_time":"1767750667"},"pure_content":"Online Auto Param Culc Software:\n\nClick Here to Tofu LensCulc\n\n\ud83d\udfe2 \u200bCustom Parameter 1 SOT Mode:\u200b\u200b\nSet to 1 for Small Target Tracking Mode, set to 0 for Standard Tracking Mode.\n\n\ud83d\udfe2 \u200bCustom Parameter 2"},{"id":"198003","string_id":"ljIpOHrE","status":"1","parent_id":"145686","slug":"NAT-Configuration","order":"4","name":"\ud83d\uddc3\ufe0fNAT Configuration","type":"2","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"","seo_keyword":"","seo_desc":"","share_switch":"0","share_code":"","author":"","published_time":"1730512008","add_time":"1730510949","update_time":"1759215399","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"2","file_type":"","auth_id":"[\"-1\"]","content":{"content_id":"198003","content":"
$\"\"$
<\/figcaption><\/figure><\/center>\n\n# NAT Configuration\n\nIntroduction: This document lists the port numbers that the device will use as a basis for port mapping.\n\n
<\/th> <\/th> <\/th><\/tr><\/thead>
Port No.<\/td> Function<\/td> Description<\/td><\/tr>
\u2705554<\/td> RTSP Video Stream<\/td> Video stream<\/td><\/tr>
\u26118000<\/td> Onvif<\/td> Onvif protocol<\/td><\/tr>
\u27058888<\/td> TVMS device control<\/td> TVMS control port<\/td><\/tr>
\u26118089<\/td> T-JSON Control<\/td> T-JSON protocol<\/td><\/tr>
\u26115555<\/td> Upgrade Port<\/td> Firmware upgrade<\/td><\/tr>
\u23f923<\/td> Debug Port<\/td> Only for Debug<\/td><\/tr>
\u23f926<\/td> Serial Server Port<\/td> Check PTZ data<\/td><\/tr><\/tbody><\/table>\n\n\u2705\uff1aMandatory\n\u2611\uff1aRecommended\n\u23f9\uff1aOptional\n
\n\n\nIf you use TVMS software, manually add the device and change the port number and video stream address after the port mapping is completed, check the 'External' option.\n![](https:\/\/resource-wangsu.helplook.net\/docker_production\/4lnqex\/article\/ljIpOHrE\/6725864e777f1.png)\n\n\n---\n
$\"\"$
<\/figcaption><\/figure><\/center>","pure_content":"NAT Configuration\nIntroduction: This document lists the port numbers that the device will use as a basis for port mapping.\n Port No. Function Description \u2705554 RTSP Video Stream Video stream \u26118000 Onvif Onvif protocol \u27058888 TVMS device control TVMS control port \u26118089 T-JSON Control T-JSON protocol \u26115555 Upgrade Port Firmware upgrade \u23f923 Debug Port Only for Debug \u23f926 Serial Server Port Check PTZ data\n\u2705\uff1aMandatory\n\u2611\uff1aRecommended\n\u23f9\uff1aOptional\n\nIf you use TVMS software, manually add the device and change the port number and video stream address after the port mapping is completed, check the 'External' option.","is_draft":"0","editor_type":"2","file_path":"","add_time":"1730510949","update_time":"1749802383"},"pure_content":"NAT Configuration\nIntroduction: This document lists the port numbers that the device will use as a basis for port mapping.\n Port No. Function Description \u2705554 RTSP Video Stream Video stream \u26118000"},{"id":"528036","string_id":"Qa2jpiaQ","status":"1","parent_id":"145686","slug":"Multispectral-Device-IP-Change-Instructions","order":"3","name":"\ud83c\udf10Multispectral Device IP Change Instructions","type":"2","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"","seo_keyword":"","seo_desc":"","share_switch":"0","share_code":"","author":"","published_time":"1749805911","add_time":"1749805510","update_time":"1759211120","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"2","file_type":"","auth_id":"[\"-1\"]","content":{"content_id":"528036","content":"
$\"\"$
<\/figcaption><\/figure><\/center>\n\n# Multispectral Device IP Change Instructions\n\nIf your device is not designed with a single IP method, please follow the procedures below to verify the IP change process. All IP addresses below must be distinct. Ensure you keep detailed modification records during the process.\n\n| Change Sequence | Application Type | Instructions |\n|------------------|----------------------------------|-------------------|\n| Visible Light Camera IP | For network camera input, confirm camera IP modification first | Change via web login |\n| Thermal Infrared Camera IP | For network camera input, confirm camera IP modification first | Change via web login |\n| Digital Interface Camera | No action required | |\n| Serial Server IP | Modify IP via web login. Default IP: 192.168.1.66
If modified previously, check Serial Server IP in TVMS client | Change via web login. Default credentials: admin\/12345678
Do not modify configurations beyond IP address |\n| Laser Fill Light\/
Laser Rangefinder
Serial Server IP | Modify IP via web login. Default IP: 192.168.1.88
If modified previously, check Laser Rangefinder IP in TVMS client | Change via web login. Default credentials: admin\/12345678
Do not modify configurations beyond IP address |\n| Tofu Module IP | 1. Modify local IP via TVMS client
2. Re-add login for new IP
3. Modify Serial Server IP configuration
4. For network camera input devices: modify stream URL
5. Update Laser Rangefinder IP | Execute steps sequentially.
Skip irrelevant device types |\n\n> Recommend using router NAT mapping or single-IP modules for IP mapping. In such cases, only modify the mapped address.\n\n---\n
$\"\"$
<\/figcaption><\/figure><\/center>","pure_content":"Multispectral Device IP Change Instructions\nIf your device is not designed with a single IP method, please follow the procedures below to verify the IP change process. All IP addresses below must be distinct. Ensure you keep detailed modification records during the process.\n\n\n\n Change Sequence\n Application Type\n Instructions\n\n\n\n\n Visible Light Camera IP\n For network camera input, confirm camera IP modification first\n Change via web login\n\n\n Thermal Infrared Camera IP\n For network camera input, confirm camera IP modification first\n Change via web login\n\n\n Digital Interface Camera\n No action required\n \n\n\n Serial Server IP\n Modify IP via web login. Default IP: 192.168.1.66If modified previously, check Serial Server IP in TVMS client\n Change via web login. Default credentials: admin\/12345678Do not modify configurations beyond IP address\n\n\n Laser Fill Light\/Laser RangefinderSerial Server IP\n Modify IP via web login. Default IP: 192.168.1.88If modified previously, check Laser Rangefinder IP in TVMS client\n Change via web login. Default credentials: admin\/12345678Do not modify configurations beyond IP address\n\n\n Tofu Module IP\n 1. Modify local IP via TVMS client2. Re-add login for new IP3. Modify Serial Server IP configuration4. For network camera input devices: modify stream URL5. Update Laser Rangefinder IP\n Execute steps sequentially.Skip irrelevant device types\n\n\n\n\nRecommend using router NAT mapping or single-IP modules for IP mapping. In such cases, only modify the mapped address.","is_draft":"0","editor_type":"2","file_path":"","add_time":"1749805510","update_time":"1749805936"},"pure_content":"Multispectral Device IP Change Instructions\nIf your device is not designed with a single IP method, please follow the procedures below to verify the IP change process. All IP addresses below must be"},{"id":"625552","string_id":"DiAXPPrw","status":"1","parent_id":"145686","slug":"Deploy-a-Custom-Model-in-Tofu5","order":"2","name":"\ud83d\ude80Deploy a Custom Model in Tofu5","type":"2","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"","seo_keyword":"","seo_desc":"","share_switch":"0","share_code":"","author":"","published_time":"1758798804","add_time":"1758798680","update_time":"1769563677","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"2","file_type":"","auth_id":"[\"-1\"]","content":{"content_id":"625552","content":"
$\"\"$
<\/figcaption><\/figure><\/center>\n\n> This function is based on the Tofu5 VMware image.\n> Recommended VMware version is VMware\u00ae Workstation 16 Pro or later version.\n> We will provide video tutorial guides in the resource package to help you quickly master the usage.\n\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\n# 1. Model Quantization\nThe model quantization method is exactly the same as that used by Ultralytics. We have modified the original code of the model export process to adapt it for model inference with the Tofu5 module.\nThe model input is at a resolution of 3x640x640, Yolov8s customized model type is supported, supports detection models with up to 4 classes.\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\nCheck your model in [Netron](https:\/\/github.com\/lutzroeder\/netron) to ensure it ends with a `detect layer`.\n\nYou can modify the `\/home\/littro\/OnnxModel_export\/export_nnie.py` file to export your own `.pt` model to an `.onnx` model\n```python\nfrom ultralytics import YOLO\nmodel = YOLO(\".\/model\/tank.pt\")\nsuccess = model.export(format=\"onnx\", opset=13)\n```\n\nAfter model quantization, You can obtain the ONNX file in the `\/home\/littro\/OnnxModel_export\/model` folder. Please use [Netron](https:\/\/github.com\/lutzroeder\/netron) to view and check the ONNX network.\nWe provide the `Netron` software in the `\/home\/littro\/Netron` folder.\nEnter the following command to run Netron, then open the model file for browsing.\n```bash\n\/home\/littro\/Netron\/Netron-.AppImage\n# Replace with the actual version number of the AppImage file\n```\nThe correct network should end as shown in the figure below.\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\n\n# 2. Deploy the Custom Model\n\n## 2.1 Environment Check\nUse the administrator account and run the following commands in sequence to check the ATC environment. \n```bash\nsudo su\n# Enter your password when prompted\natc --version\n```\nIf the information below is printed, the environment is normal.\n```bash\nMapper Version \n```\n\n## 2.2 Prepare Files\n\n1. `.onnx file` (converted from a `.pt` model). We have provided a reference `.onnx` file located in `\/home\/littro\/Model\/yolov8\/`.\u200b\n2. `Reference images and a .txt image list file`. We have provided reference images and a `.txt` file located in `\/home\/littro\/Model\/coco\/`.\u200b\n3. A folder for storing the exported model (create it in advance if it does not exist).\n\n## 2.3 Export Model\n\nUse the administrator account and run the following commands to switch to administrator mode.\n```bash\nsudo su\n# Enter your password when prompted\n\n```\n\nRun the code after modifying the content.\n\n\ud83d\udfe2 Reference command:\n```bash\natc --dump_data=0 --input_shape=\"images:1,3,640,640\" \\\n--input_type=\"images:UINT8\" --log_level=0 --online_model_type=0 --batch_num=1 \\\n--input_format=NCHW --output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\" \\\n--soc_version=Hi3516DV500 \\\n--insert_op_conf=\/root\/modelzoo\/tank\/Hi3516DV500\/insert_op.cfg \\\n--framework=5 --compile_mode=0 --save_original_model=false \\\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\" \\\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\" \n```\n\nSections that need to be modified manually:\n\n\ud83d\udfe2 output: The output path.\nThe refence in the code is:\n```bash\n--output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\"\n```\nOutput `.om` model will be named `tank.om` located in the `\/home\/littro\/Model\/yolov8\/om_model\/` foulder.\n\n\ud83d\udfe2 model: The path and name of the ONNX model.\nThe refence in the code is:\n```bash\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\"\n```\n\n\ud83d\udfe2 image_list: Path to the .txt file containing the list of reference images for model generation. The list must include 20 to 50 images from the training dataset (ensure the images are consistent with the model's training data distribution to avoid quantization errors).\nThe refence in the code is:\n```bash\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\"\n```\n\nThe image list is saved as a`.txt`file,with the format as follows:\n ```text\n\/home\/littro\/Model\/coco\/000000000036.jpg\n\/home\/littro\/Model\/coco\/000000000049.jpg\n\/home\/littro\/Model\/coco\/000000000061.jpg\n\/home\/littro\/Model\/coco\/000000000074.jpg\n\/home\/littro\/Model\/coco\/000000000077.jpg\n\/home\/littro\/Model\/coco\/000000000086.jpg\n\/home\/littro\/Model\/coco\/000000000109.jpg\n\/home\/littro\/Model\/coco\/000000000110.jpg\n```\n\nWe provided the reference images and `.txt` file located in `\/home\/littro\/Model\/coco\/`.\nThe model export process will take several minutes.\nOnce the export is successful, the resulting file will be in the `.om` format.\n\n## 2.4 Model Packaging\nRun the packaging script.\nFirst, switch to the script directory (path: \/home\/littro\/Model\/yolov8\/), then run the packaging script\n```bash\n# Switch to the script directory\ncd \/home\/littro\/Model\/yolov8\/\n# Run the packaging script\n.\/encryptmodel tank.om model_custom.ir\n```\nAfter running, the packaged model file named `model_custom.ir` will be obtained.\n\n## 2.5 Create an Upgrade Package\nScript path:`\/home\/littro\/Model\/updpkg_custom_script`\n\nCopy the generated `model_custom.ir` model to the specified path and run the script.\n```bash\n.\/updpkg_custom_model_vis.sh .\/model_custom.ir\n```\n\n>Please note that the tank.om model is generated under the administrator account. If you copy the file to the target directory using a regular user, run the following command to grant read\/write\/execute permissions (replace file_name with the actual file name, e.g., model_custom.ir):\n>```bash\n>chmod 777 file_name\n>```\n\nAfter running,the `.dev` upgrade package will be generated automatically.\nThis upgrade package can be used for remote upgrades on Tofu5.\n\n# 3. Model Upgrades and Testing\n\nFor detailed upgrade procedures, please refer to the TVMS software user manual[[click here]](https:\/\/tofuai.helplook.net\/docs\/tvms#2.2.5_Device_Software_Upgrade).\n\nAfter the upgrade is completed, enter the Tofu5 device settings interface, find the AI Model Type option, and change it to `Customize Detection`.\n\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\nThe result of the demo model is shown below; it can detect tanks in the video.\n\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\n---\n
$\"\"$
<\/figcaption><\/figure><\/center>","pure_content":"This function is based on the Tofu5 VMware image.\nRecommended VMware version is VMware\u00ae Workstation 16 Pro or later version.\nWe will provide video tutorial guides in the resource package to help you quickly master the usage.\n\n\n1. Model Quantization\nThe model quantization method is exactly the same as that used by Ultralytics. We have modified the original code of the model export process to adapt it for model inference with the Tofu5 module.\nThe model input is at a resolution of 3x640x640, Yolov8s customized model type is supported, supports detection models with up to 4 classes.\n\nCheck your model in Netron to ensure it ends with a detect layer.\nYou can modify the \/home\/littro\/OnnxModel_export\/export_nnie.py file to export your own .pt model to an .onnx model\nfrom ultralytics import YOLO\nmodel = YOLO(\".\/model\/tank.pt\")\nsuccess = model.export(format=\"onnx\", opset=13)\n\nAfter model quantization, You can obtain the ONNX file in the \/home\/littro\/OnnxModel_export\/model folder. Please use Netron to view and check the ONNX network.\nWe provide the Netron software in the \/home\/littro\/Netron folder.\nEnter the following command to run Netron, then open the model file for browsing.\n\/home\/littro\/Netron\/Netron-.AppImage\n# Replace * with the actual version number of the AppImage file\n\nThe correct network should end as shown in the figure below.\n\n2. Deploy the Custom Model\n2.1 Environment Check\nUse the administrator account and run the following commands in sequence to check the ATC environment.\nsudo su\n# Enter your password when prompted\natc --version\n\nIf the information below is printed, the environment is normal.\nMapper Version \n\n2.2 Prepare Files\n\n\n.onnx file (converted from a .pt model). We have provided a reference .onnx file located in \/home\/littro\/Model\/yolov8\/.\u200b\n\nReference images and a .txt image list file. We have provided reference images and a .txt file located in \/home\/littro\/Model\/coco\/.\u200b\nA folder for storing the exported model (create it in advance if it does not exist).\n\n2.3 Export Model\nUse the administrator account and run the following commands to switch to administrator mode.\nsudo su\n# Enter your password when prompted\n\n\nRun the code after modifying the content.\n\ud83d\udfe2 Reference command:\natc --dump_data=0 --input_shape=\"images:1,3,640,640\" \\\n--input_type=\"images:UINT8\" --log_level=0 --online_model_type=0 --batch_num=1 \\\n--input_format=NCHW --output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\" \\\n--soc_version=Hi3516DV500 \\\n--insert_op_conf=\/root\/modelzoo\/tank\/Hi3516DV500\/insert_op.cfg \\\n--framework=5 --compile_mode=0 --save_original_model=false \\\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\" \\\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\" \n\nSections that need to be modified manually:\n\ud83d\udfe2 output: The output path.\nThe refence in the code is:\n--output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\"\n\nOutput .om model will be named tank.om located in the \/home\/littro\/Model\/yolov8\/om_model\/ foulder.\n\ud83d\udfe2 model: The path and name of the ONNX model.\nThe refence in the code is:\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\"\n\n\ud83d\udfe2 image_list: Path to the .txt file containing the list of reference images for model generation. The list must include 20 to 50 images from the training dataset (ensure the images are consistent with the model's training data distribution to avoid quantization errors).\nThe refence in the code is:\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\"\n\nThe image list is saved as a.txtfile,with the format as follows:\n\/home\/littro\/Model\/coco\/000000000036.jpg\n\/home\/littro\/Model\/coco\/000000000049.jpg\n\/home\/littro\/Model\/coco\/000000000061.jpg\n\/home\/littro\/Model\/coco\/000000000074.jpg\n\/home\/littro\/Model\/coco\/000000000077.jpg\n\/home\/littro\/Model\/coco\/000000000086.jpg\n\/home\/littro\/Model\/coco\/000000000109.jpg\n\/home\/littro\/Model\/coco\/000000000110.jpg\n\nWe provided the reference images and .txt file located in \/home\/littro\/Model\/coco\/.\nThe model export process will take several minutes.\nOnce the export is successful, the resulting file will be in the .om format.\n2.4 Model Packaging\nRun the packaging script.\nFirst, switch to the script directory (path: \/home\/littro\/Model\/yolov8\/), then run the packaging script\n# Switch to the script directory\ncd \/home\/littro\/Model\/yolov8\/\n# Run the packaging script\n.\/encryptmodel tank.om model_custom.ir\n\nAfter running, the packaged model file named model_custom.ir will be obtained.\n2.5 Create an Upgrade Package\nScript path:\/home\/littro\/Model\/updpkg_custom_script\nCopy the generated model_custom.ir model to the specified path and run the script.\n.\/updpkg_custom_model_vis.sh .\/model_custom.ir\n\n\nPlease note that the tank.om model is generated under the administrator account. If you copy the file to the target directory using a regular user, run the following command to grant read\/write\/execute permissions (replace file_name with the actual file name, e.g., model_custom.ir):\nchmod 777 file_name\n\n\nAfter running,the .dev upgrade package will be generated automatically.\nThis upgrade package can be used for remote upgrades on Tofu5.\n3. Model Upgrades and Testing\nFor detailed upgrade procedures, please refer to the TVMS software user manual[click here].\nAfter the upgrade is completed, enter the Tofu5 device settings interface, find the AI Model Type option, and change it to Customize Detection.\n\nThe result of the demo model is shown below; it can detect tanks in the video.","is_draft":"0","editor_type":"2","file_path":"","add_time":"1758798680","update_time":"1769563677"},"pure_content":"This function is based on the Tofu5 VMware image.\nRecommended VMware version is VMware\u00ae Workstation 16 Pro or later version.\nWe will provide video tutorial guides in the resource package to help you"},{"id":"577235","string_id":"xxwaCM1j","status":"1","parent_id":"145686","slug":"deploy_om","order":"1","name":"\ud83d\ude80Deploy a Custom Model in Tofu6","type":"2","cover":"","cover_alt":"","desc":null,"website":"","contact":"","seo_title":"Deploy your own model using Tofu6","seo_keyword":"","seo_desc":"","share_switch":"0","share_code":"","author":"","published_time":"1753663939","add_time":"1753663891","update_time":"1769563686","corp_auth_id":"0","verify_code_auth_id":"0","editor_type":"2","file_type":"","auth_id":"[\"-1\"]","content":{"content_id":"577235","content":"
$\"\"$
<\/figcaption><\/figure><\/center>\n\n> This function is based on the Tofu6 VMware image.\n> Recommended VMware version is VMware\u00ae Workstation 16 Pro or later version.\n> We will provide video tutorial guides in the resource package to help you quickly master the usage.\n\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\n# 1. Model Quantization\nThe model quantization method is exactly the same as that used by Ultralytics. We have modified the original code of the model export process to adapt it for model inference with the Tofu6 module.\nThe model input is at a resolution of 3x640x640, Yolov8s customized model type is supported , supports detection models with up to 4 classes.\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\nCheck your model in [Netron](https:\/\/github.com\/lutzroeder\/netron) to ensure it ends with a `detect layer`.\n\nYou can modify the `\/home\/littro\/OnnxModel_export\/export_nnie.py` file to export your own `.pt` model to an `.onnx` model\n```python\nfrom ultralytics import YOLO\nmodel = YOLO(\".\/model\/tank.pt\")\nsuccess = model.export(format=\"onnx\", opset=13)\n```\n\nAfter model quantization, You can obtain the ONNX file in the `\/home\/littro\/OnnxModel_export\/model` folder. Please use [Netron](https:\/\/github.com\/lutzroeder\/netron) to view and check the ONNX network.\nWe provide the `Netron` software in the `\/home\/littro\/Netron` folder.\nEnter the following command to run Netron, then open the model file for browsing.\n```bash\n\/home\/littro\/Netron\/Netron-.AppImage\n# Replace * with the actual version number of the AppImage file\n```\nThe correct network should end as shown in the figure below.\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\n\n# 2. Deploy the Custom Model\n\n## 2.1 Environment Check\nUse the administrator account and run the following commands in sequence to check the ATC environment. \n```bash\nsudo su\n# Enter your password when prompted\natc --version\n```\nIf the information below is printed, the environment is normal.\n```bash\nMapper Version \n```\n\n## 2.2 Prepare Files\n\n1. `.onnx file` (converted from a `.pt` model). We have provided a reference `.onnx` file located in `\/home\/littro\/Model\/yolov8\/`.\u200b\n2. `Reference images and a .txt image list file`. We have provided reference images and a `.txt` file located in `\/home\/littro\/Model\/coco\/`.\u200b\n3. A folder for storing the exported model (create it in advance if it does not exist).\n\n## 2.3 Export Model\n\nUse the administrator account and run the following commands to switch to administrator mode.\n```bash\nsudo su\n# Enter your password when prompted\n\n```\n\nRun the code after modifying the content.\n\n\ud83d\udfe2 Reference command:\n```bash\natc --dump_data=0 --input_shape=\"images:1,3,640,640\" \\\n--input_type=\"images:UINT8\" --log_level=0 --online_model_type=0 --batch_num=1 \\\n--input_format=NCHW \\\n--output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\" \\\n--soc_version=SS928V100 \\\n--insert_op_conf=\/root\/modelzoo\/yolov8s\/SS928V100\/insert_op.cfg \\\n--framework=5 --compile_mode=0 --save_original_model=false \\\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\" \\\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\" \n```\n\nSections that need to be modified manually:\n\n\ud83d\udfe2 output: The output path.\nThe refence in the code is:\n```bash\n--output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\"\n```\nOutput `.om` model will be named `tank.om` located in the `\/home\/littro\/Model\/yolov8\/om_model\/` foulder.\n\n\ud83d\udfe2 model: The path and name of the ONNX model.\nThe refence in the code is:\n```bash\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\"\n```\n\n\ud83d\udfe2 image_list: Path to the .txt file containing the list of reference images for model generation. The list must include 20 to 50 images from the training dataset (ensure the images are consistent with the model's training data distribution to avoid quantization errors).\nThe refence in the code is:\n```bash\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\"\n```\n\nThe image list is saved as a`.txt`file,with the format as follows:\n ```text\n\/home\/littro\/Model\/coco\/000000000036.jpg\n\/home\/littro\/Model\/coco\/000000000049.jpg\n\/home\/littro\/Model\/coco\/000000000061.jpg\n\/home\/littro\/Model\/coco\/000000000074.jpg\n\/home\/littro\/Model\/coco\/000000000077.jpg\n\/home\/littro\/Model\/coco\/000000000086.jpg\n\/home\/littro\/Model\/coco\/000000000109.jpg\n\/home\/littro\/Model\/coco\/000000000110.jpg\n```\n\nWe provided the reference images and `.txt` file located in `\/home\/littro\/Model\/coco\/`.\nThe model export process will take several minutes.\nOnce the export is successful, the resulting file will be in the `.om` format.\n\n## 2.4 Model Packaging\nRun the packaging script.\nFirst, switch to the script directory (path: \/home\/littro\/Model\/yolov8\/), then run the packaging script\n```bash\n# Switch to the script directory\ncd \/home\/littro\/Model\/yolov8\/\n# Run the packaging script\n.\/encryptmodel tank.om model_custom.ir\n```\nAfter running, the packaged model file named `model_custom.ir` will be obtained.\n\n## 2.5 Create an Upgrade Package\nScript path:`\/home\/littro\/Model\/updpkg_custom_script`\n\nCopy the generated `model_custom.ir` model to the specified path and run the script.\n```bash\n.\/updpkg_custom_model_vis.sh .\/model_custom.ir\n```\n\n>Please note that the tank.om model is generated under the administrator account. If you copy the file to the target directory using a regular user, run the following command to grant read\/write\/execute permissions (replace file_name with the actual file name, e.g., model_custom.ir):\n>```bash\n>chmod 777 file_name\n>```\n\nAfter running,the `.dev` upgrade package will be generated automatically.\nThis upgrade package can be used for remote upgrades on Tofu6.\n\n# 3. Model Upgrades and Testing\n\nFor detailed upgrade procedures, please refer to the TVMS software user manual[[click here]](https:\/\/tofuai.helplook.net\/docs\/tvms#2.2.5_Device_Software_Upgrade).\n\nAfter the upgrade is completed, enter the Tofu6 device settings interface, find the AI Model Type option, and change it to `Customize Detection`.\n\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\nThe result of the demo model is shown below; it can detect tanks in the video.\n\n
$\"\"$
<\/figcaption><\/figure><\/center>\n\n---\n
$\"\"$
<\/figcaption><\/figure><\/center>","pure_content":"This function is based on the Tofu6 VMware image.\nRecommended VMware version is VMware\u00ae Workstation 16 Pro or later version.\nWe will provide video tutorial guides in the resource package to help you quickly master the usage.\n\n\n1. Model Quantization\nThe model quantization method is exactly the same as that used by Ultralytics. We have modified the original code of the model export process to adapt it for model inference with the Tofu6 module.\nThe model input is at a resolution of 3x640x640, Yolov8s customized model type is supported , supports detection models with up to 4 classes.\n\nCheck your model in Netron to ensure it ends with a detect layer.\nYou can modify the \/home\/littro\/OnnxModel_export\/export_nnie.py file to export your own .pt model to an .onnx model\nfrom ultralytics import YOLO\nmodel = YOLO(\".\/model\/tank.pt\")\nsuccess = model.export(format=\"onnx\", opset=13)\n\nAfter model quantization, You can obtain the ONNX file in the \/home\/littro\/OnnxModel_export\/model folder. Please use Netron to view and check the ONNX network.\nWe provide the Netron software in the \/home\/littro\/Netron folder.\nEnter the following command to run Netron, then open the model file for browsing.\n\/home\/littro\/Netron\/Netron-.AppImage\n# Replace * with the actual version number of the AppImage file\n\nThe correct network should end as shown in the figure below.\n\n2. Deploy the Custom Model\n2.1 Environment Check\nUse the administrator account and run the following commands in sequence to check the ATC environment.\nsudo su\n# Enter your password when prompted\natc --version\n\nIf the information below is printed, the environment is normal.\nMapper Version *\n\n2.2 Prepare Files\n\n\n.onnx file (converted from a .pt model). We have provided a reference .onnx file located in \/home\/littro\/Model\/yolov8\/.\u200b\n\nReference images and a .txt image list file. We have provided reference images and a .txt file located in \/home\/littro\/Model\/coco\/.\u200b\nA folder for storing the exported model (create it in advance if it does not exist).\n\n2.3 Export Model\nUse the administrator account and run the following commands to switch to administrator mode.\nsudo su\n# Enter your password when prompted\n\n\nRun the code after modifying the content.\n\ud83d\udfe2 Reference command:\natc --dump_data=0 --input_shape=\"images:1,3,640,640\" \\\n--input_type=\"images:UINT8\" --log_level=0 --online_model_type=0 --batch_num=1 \\\n--input_format=NCHW \\\n--output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\" \\\n--soc_version=SS928V100 \\\n--insert_op_conf=\/root\/modelzoo\/yolov8s\/SS928V100\/insert_op.cfg \\\n--framework=5 --compile_mode=0 --save_original_model=false \\\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\" \\\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\" \n\nSections that need to be modified manually:\n\ud83d\udfe2 output: The output path.\nThe refence in the code is:\n--output=\"\/home\/littro\/Model\/yolov8\/om_model\/tank\"\n\nOutput .om model will be named tank.om located in the \/home\/littro\/Model\/yolov8\/om_model\/ foulder.\n\ud83d\udfe2 model: The path and name of the ONNX model.\nThe refence in the code is:\n--model=\"\/home\/littro\/Model\/yolov8\/tank.onnx\"\n\n\ud83d\udfe2 image_list: Path to the .txt file containing the list of reference images for model generation. The list must include 20 to 50 images from the training dataset (ensure the images are consistent with the model's training data distribution to avoid quantization errors).\nThe refence in the code is:\n--image_list=\"images:\/home\/littro\/Model\/coco\/imagelist_20240611134217.txt\"\n\nThe image list is saved as a.txtfile,with the format as follows:\n\/home\/littro\/Model\/coco\/000000000036.jpg\n\/home\/littro\/Model\/coco\/000000000049.jpg\n\/home\/littro\/Model\/coco\/000000000061.jpg\n\/home\/littro\/Model\/coco\/000000000074.jpg\n\/home\/littro\/Model\/coco\/000000000077.jpg\n\/home\/littro\/Model\/coco\/000000000086.jpg\n\/home\/littro\/Model\/coco\/000000000109.jpg\n\/home\/littro\/Model\/coco\/000000000110.jpg\n\nWe provided the reference images and .txt file located in \/home\/littro\/Model\/coco\/.\nThe model export process will take several minutes.\nOnce the export is successful, the resulting file will be in the .om format.\n2.4 Model Packaging\nRun the packaging script.\nFirst, switch to the script directory (path: \/home\/littro\/Model\/yolov8\/), then run the packaging script\n# Switch to the script directory\ncd \/home\/littro\/Model\/yolov8\/\n# Run the packaging script\n.\/encryptmodel tank.om model_custom.ir\n\nAfter running, the packaged model file named model_custom.ir will be obtained.\n2.5 Create an Upgrade Package\nScript path:\/home\/littro\/Model\/updpkg_custom_script\nCopy the generated model_custom.ir model to the specified path and run the script.\n.\/updpkg_custom_model_vis.sh .\/model_custom.ir\n\n\nPlease note that the tank.om model is generated under the administrator account. If you copy the file to the target directory using a regular user, run the following command to grant read\/write\/execute permissions (replace file_name with the actual file name, e.g., model_custom.ir):\nchmod 777 file_name\n\n\nAfter running,the .dev upgrade package will be generated automatically.\nThis upgrade package can be used for remote upgrades on Tofu6.\n3. Model Upgrades and Testing\nFor detailed upgrade procedures, please refer to the TVMS software user manual[click here].\nAfter the upgrade is completed, enter the Tofu6 device settings interface, find the AI Model Type option, and change it to Customize Detection.\n\nThe result of the demo model is shown below; it can detect tanks in the video.","is_draft":"0","editor_type":"2","file_path":"","add_time":"1753663891","update_time":"1769563685"},"pure_content":"This function is based on the Tofu6 VMware image.\nRecommended VMware version is VMware\u00ae Workstation 16 Pro or later version.\nWe will provide video tutorial guides in the resource package to help you"}],"content":{"icon":"\/docker_production\/4lnqex\/content_icon\/MT62dV?rand=195242466","desc":"","content_id":"145686","add_time":"1724144130","update_time":"1749802414","content":"","is_traffic_less":"0","file_type":""},"hot_list":[],"user_read_num":"0","is_auth":"0"},"msg":"Success","code":200}

<\/h2>\n

2. AI Module Interface<\/h2>\n

Ethernet connector<\/h3>\nWe provide waterproof kits, connect the network cable as shown in the figure.<\/p>\n<\/p>\n<\/p>\n <\/p>\n

Power In\/Out Connector<\/h3>\nThe specification of the power connector is 5525, and the connection method is shown in the following figure.<\/p>\n<\/p>\n<\/p>\n

$\"\"$ <\/h2>\n

Ethernet connector<\/h3>\n
We provide waterproof kits, connect the network cable as shown in the figure.<\/p>\n
<\/p>\n
<\/p>\n
<\/p>\n

Power In\/Out Connector<\/h3>\n
The specification of the power connector is 5525, and the connection method is shown in the following figure.<\/p>\n
<\/p>\n
<\/p>\n