Buckeye X Series User Manual

Network manager ai object detection

Hide thumbs

Table Of Contents

Table of Contents

Quick Links

Download this manual

X-Series Network Manager AI Object Detection

User's Manual

AI Object Detection User's Manual Version 1.0

Table of Contents

Need help?

Do you have a question about the X Series and is the answer not in the manual?

Questions and answers

Summary of Contents for Buckeye X Series

Page 1 X-Series Network Manager AI Object Detection User's Manual AI Object Detection User's Manual Version 1.0...
Page 2: Table Of Contents
Table of Contents 1.Introduction.............................1 2.Installation...............................3 2.1.Computer Requirements........................3 2.2.How To Install ..........................3 2.2.1.Downloading base model through software................3 2.2.2.Download from our website ....................5 3.AI Detection profiles ..........................6 3.1.Navigating to Detection Profiles ....................6 3.2.Customizing AI Detection Profiles ....................10 3.3.Profile Information ........................11 3.3.1.Profile Name .........................11 3.3.2.Profile Active.........................11 3.3.3.Profile applies to cameras .....................11 3.4.Target Specific Information......................12...
Page 3: Introduction
Vehicle • Additional targets of interest may be added in the future as the AI-OD models improve. Contact ATSI/Buckeye Cam service at www.buckeyecam.com for specialized target requests. If a target of interest is detected in an image, the software can take various user defined actions such as: Email images to a special email address •...
Page 4 1. Camera detects motion and takes a picture. This picture then gets transmitted to the base. 2. Once the image has arrived at the base, we look to see if the camera is a part of any active detection profiles. If so, we run the image through the AI process that will return detected targets if any are detected.
Page 5: Installation
2. Installation 2.1. Computer Requirements Minimum requirements are: Memory: 8GB RAM or higher • Processor: Core i3-3225 3.3 GHz or higher • These are the minimum requirements and users may experience performance delays with minimum hardware configuration. Newer laptops and computers should run our AI. High quality processors and graphics cards can reduce time it takes to process each image.
Page 6 Next click the check for updates button,shown in the red box. AI Object Detection User's Manual Version 1.0 Page 4 of 32...
Page 7: Download From Our Website
Then click the “Download AI” button. Follow the prompt to download the base AI model. NOTE: This will only appear if the software determines that your computer is capable of running the AI software. 2.2.2. Download from our website Specific models can be downloaded from our website. For more information contact tech support at www.buckeyecam.com AI Object Detection User's Manual Version 1.0 Page 5 of 32...
Page 8: Ai Detection Profiles
3. AI Detection profiles An AI detection profile is a user defined profile that controls what actions the X-Series Manager needs to take when an image comes back with detected targets. Detection profiles can define things like: Where to email images on detection of targets •...
Page 9 Next, click on the “Notifications” button shown in the red box below. AI Object Detection User's Manual Version 1.0 Page 7 of 32...
Page 10 Then, scroll down and click the edit button in the “AI Object Detection” Tab shown in the red box below. AI Object Detection User's Manual Version 1.0 Page 8 of 32...
Page 11 You will now be on this screen. This screen is where you can add, remove, or modify your AI detection profiles. To add a profile, click the add detection profile button shown in the red box below. AI Object Detection User's Manual Version 1.0 Page 9 of 32...
Page 12: Customizing Ai Detection Profiles
3.2. Customizing AI Detection Profiles After clicking the add profile button, this screen will appear. Let's walk-through some of the features you can customize on this profile. AI Object Detection User's Manual Version 1.0 Page 10 of 32...
Page 13: Profile Information
3.3. Profile Information The first section is for information that applies to the overall profile. This includes: 3.3.1. Profile Name This field allows the user to enter a custom name for their detection profile. For example, if this profile is designed to catch vehicles on a driveway, you could name it “Driveway” to quickly reference the profile after completion.
Page 14: Target Specific Information
3.4. Target Specific Information The second section is your target information. Targets are what the particular model you have selected can detect. Examples of targets are: People • Vehicles • For example, you may want to have a different action when you detect a person in a location vs a vehicle.
Page 15: Email Pictures
3.4.3. Email Pictures This feature allows you to email any picture that comes back as a hit on the specific target to an email address of your choosing. By checking the checkbox “List detected targets in subject”,the target that was detected will automatically be put in the subject of the email. After setting up your outgoing email server, enter the email address you would like to send in the white text box.
Page 16: Speak
3.4.5. Speak This feature allows you to have a custom sound played on your computer system when a picture comes back as a detection on the specific target. For example, in a home security application if you type in “Intruder” your computer would say “Intruder” through its system's audio quickly alerting you of the situation.
Page 17 AI Object Detection User's Manual Version 1.0 Page 15 of 32...
Page 18: Example Profiles
4. Example Profiles Example 1: John is an owner of a construction company. He expects daytime traffic at his work sight, but wants to be notified if his camera picks up any motion at night on the driveway leading to work sight. He created a profile called Driveways and set it to active so it will be running.
Page 19 AI Object Detection User's Manual Version 1.0 Page 17 of 32...
Page 20 Example 2: Bob is monitoring a dirt road for foot traffic. In this area it is very uncommon for there to be foot traffic unless there is illegal activity taking place. Bob wants “person” target detections to be sent to one email that he can actively monitor, and vehicle target detections to be sent to other email for later inspection, and all other images to not get sent to an email.
Page 21 AI Object Detection User's Manual Version 1.0 Page 19 of 32...
Page 22 Here is an example of the Detection Profiles Screen after creating those profiles. AI Object Detection User's Manual Version 1.0 Page 20 of 32...
Page 23: Detection Zones
Detection Zones 5.1. Introduction to detection zones Detection zones are available to allow certain areas of an image to be analyzed, suppressed, or ignored. They DO NOT need to be set up for AI to work and should only be used in specific situations. There are three main types of detection zones 1.
Page 24: Navigating To Detection Zones
5.3. Navigating to detection zones To navigate to the detection zones, first click on the camera icon in the device tree that you want to modify. The screen below should appear. Click the Detection Zones button(shown in red box) to modify the detection zones.
Page 25 After Clicking the “Detection Zone” button, a screen similar to the one below will appear In this screen you can add areas of suppression or areas of interest. The info icons, shown in blue boxes, can give you more information on each topic. AI Object Detection User's Manual Version 1.0 Page 23 of 32...
Page 26: Zones Of Interest
5.4. Zones of Interest Zones of interest are areas in the images you are specifically interested in getting a target detection from. IMPORTANT: Any target detection that has at least one part of its bounding box touching at least on part of the zone of interest will be kept.
Page 27 In this example Bob wants to watch the dumpster at his worksite because he thinks people are illegally putting their trash in his dumpster. He does not want to trigger on the public road in the top of the image,therefore, he places an area of interest around the dumpster. This is how this might look. As you can see the purple bounding box of the person touches the zone of interest so it is kept.
Page 28: Suppressing Targets
5.5. Suppressing Targets When specific areas in an image cause repeated incorrect target detections, they are good candidates to suppress. An example of this is a strange shaped bush that repeatedly gets detected as a person, or an old rusty truck cab in a field that repeatedly gets identified as a vehicle. Suppressing targets is done automatically for you when you click the red arrow to the right of the detected target you want to suppress.
Page 29 In this example Bob wants to watch a driveway for incoming traffic. There is a stationary van that is broke down and never moves. Bob suppresses this target to allow the white vehicle to be correctly be detected even though its bounding box touches the suppressed target. AI Object Detection User's Manual Version 1.0 Page 27 of 32...
Page 30: Zones Of Suppression
5.6. Zones of Suppression Zones of suppression are areas where the AI will ignore target detections. They can be used if there are elements in your image that are repeatably being identified as a target, but you would prefer them not to be.
Page 31 In the example below Bob wants to detect people approaching his back deck. He is not concerned with people that are detected in the grass, therefore he creates a zone of suppression to ignore target detections in the grass. A more advanced example of combining zone of interest and zone of suppression is shown below. AI Object Detection User's Manual Version 1.0 Page 29 of 32...
Page 32 In this example Bob is monitoring a dead end road that leads to his house. He only wants to be notified of vehicles on the road, not the driveway. Due to hiding locations of the camera, he had to include the driveway in the images.