By Marco Fontani, Forensic Director at Amped Software
One common trait of crimes and incidents captured by surveillance cameras is that they happen very quickly. Although no scientific evidence for this claim is easily found (may be worth some research!), it’s quite common to have car incidents, robberies, or shootings that happen in a matter of seconds.
The problem is when you’re given a few hours of footage to inspect, it’s not always possible to know in advance where to look. Finding those few moments of interest can take a long time if you’re left alone with a player. Not to mention the fact that playing surveillance footage is a challenge in itself, given the extensive amount of proprietary video formats out there.
Luckily, a solution to both problems exists—it’s called Amped Replay. Replay is the enhanced video player for investigators and frontline officers. It embeds the powerful Amped video conversion engine, a frame-accurate player, basic correction tools to manage interlacing, aspect ratio, brightness, and contrast, along with a powerful annotation and redaction toolkit. All of this is conveniently organized in a workflow suggested by the tabs at the top of the interface.
When confronted with a proprietary video, just drag the file into Replay’s GUI. Conversion, including possible camera demultiplexing and extraction of timestamps, will automatically run, and the resulting video stream(s) will be loaded.
So now we have our video ready to be reviewed. Replay lets you quickly scroll through the video just by dragging the red marker in the player bar, but if you have hours of recordings, skimming through them without losing anything is challenging. That’s where Replay’s novel Motion Detection feature comes so in handy.
The Motion Detection Panel
The new Motion Detection panel is in Replay’s Play tab. The buttons in the panel let you run and configure the analysis, locating and possibly bookmarking motion events in your video.
Before moving on, it’s important to know that Replay’s Motion Detection works by estimating the static background of the scene and detecting changes. Therefore, Motion Detection will be useful with long CCTV clips, where the camera is static. Motion Detection will not be as helpful on a shaky video captured by a bodycam, dashcam, or handheld smartphone, simply because everything moves there!
Running the Motion Detection Analysis
To initiate motion analysis, select the first button in the Motion Detection panel, which is represented by a “running man” icon. This action begins the analysis of the video clip, monitored by a progress bar.
Upon completion, red blocks will overlay the areas where motion has been detected. The blocks are semi-transparent to allow the visual content behind them to be seen. For instance, in the example provided, the two moving subjects—marked by blue arrows—are accurately highlighted as they walk through the frame. In contrast, two individuals standing on the pavement in the background (highlighted by a yellow square) remain undetected, as they are stationary in this frame. Furthermore, the movement of the male subject’s right arm in the bottom right corner (indicated by a green arrow) is effectively captured by the tool.
How does Amped Replay achieve this? Unlike neural network-based detection systems, Amped Replay employs a pixel-based comparison technique. This process involves examining sequential frames of the video and flagging areas where a significant number of pixels change from one frame to the next. When such pixel variation is identified, the tool registers it as a motion event and overlays it for easy identification. This approach is computationally efficient and ensures forensic accuracy by relying solely on numerical pixel pattern comparisons between consecutive frames.
Once the analysis is completed, users can toggle the red overlay display as needed. To remove or restore the overlay, simply click the same button used to initiate the analysis. This control is useful when revisiting saved projects, as rerunning the analysis is not generally necessary; motion data for each frame is saved within the project file. However, if you wish to view again the overlay or modify the selected region of interest, you will be asked to rerun the analysis.
Defining a Region of Interest
Amped Replay allows the selection of a specific Region of Interest (ROI) for motion detection analysis. This feature is particularly useful in scenarios where a subject of interest or vehicle is expected to appear in a specific area of the video. For instance, an eyewitness might report seeing a suspect exit the crime scene from a particular footpath, or a suspect vehicle might have been spotted turning at a well-known intersection captured in CCTV footage.
To speed up the analysis and reduce false positives, you can draw a rectangular selection of any size or position within the video. The motion detection algorithm will then restrict its analysis to the designated ROI. Note, however, that a full analysis of all pixels must be completed before an ROI can be specified.
To define an ROI, click on the second button in the Motion Detection panel, represented by a dotted rectangular icon. Then, click and hold the mouse on the viewer, dragging it to create a rectangular selection of your desired size and position. Once you are satisfied, simply release the mouse to confirm your selection.
The ROI will appear as a solid magenta outline over the video, and you can re-draw it as many times as necessary until the desired area is isolated. Changes to the ROI are quickly processed as the motion analysis is only run once at the beginning, and stored motion data is simply re-interpreted based on the new ROI. Currently, only one ROI can be set at a time; its coordinates are displayed in the Motion Detection panel and will also be included in any generated report.
To reset the analysis to cover the entire frame, click the button with a crossed rectangular dotted icon.
This targeted approach enhances the precision of motion detection analysis, allowing investigators to concentrate on high-priority areas within video footage.
Navigating and Bookmarking Motion Events
After running the analysis and possibly setting an ROI, how can we efficiently access and mark the detected motion events? The Motion Detection panel offers straightforward navigation: use the “Previous Event” and “Next Event” buttons, represented by left and right arrow icons, to move sequentially through each detected event. Each event is tagged with a starting and ending frame, allowing quick reference and review.
For a more visual approach, bookmarking motion events provides an at-a-glance view within the timeline. Motion detection bookmarks appear in magenta, making them easily distinguishable from standard bookmarks you may have added previously. This color-coded system allows users to see precisely where motion was detected within the video without manually scrolling through frames.
Importantly, bookmarks are automatically included in Amped Replay’s generated reports, where they are displayed as frame numbers and accompanied by visual illustrations. If needed, all bookmarks can be exported as images from the Export tab for detailed review or documentation.
Understanding and Utilizing the Motion Plot
The Motion Plot provides a graphical representation of motion intensity throughout your video, allowing for quick identification of timeline sections with high or low-motion activity. This feature becomes available once you have run a motion detection analysis, as outlined previously. To activate the Motion Plot, simply click the Motion Plot button located at the bottom of the Amped Replay interface; this button is accessible across the Play, Enhance, Annotate, and Export tabs.
Once activated, the Motion Plot appears on your timeline, with high peaks indicating frames containing significant motion and lower peaks representing frames with minimal or no motion. This visual layout functions similarly to the audio plot, offering the same zoom-in capabilities. The ability to zoom over part of the plot is very important for long videos, where some events could be very short in time to be seen at full scale.
The Motion Plot interface includes two main bars: the bottom bar displays the entire motion plot for the video, while the top bar provides a zoomed-in view of a specific section. To zoom in or out on the top bar, simply hover over it and use your mouse’s scroll wheel. Navigation remains flexible, allowing you to scroll through all video frames by dragging the playhead left or right as needed.
Additionally, portions of the video where motion surpasses a defined threshold will be highlighted on the top bar, offering an at-a-glance reference to areas of interest—an aspect we’ll delve into in the following section.
Customizing the Motion Threshold
To focus only on motion events relevant to your investigation, you may need to set a motion threshold in addition to or instead of defining a Region of Interest (ROI). This can be particularly useful when dealing with videos containing frequent motion events, such as those caused by traffic, objects moved by wind, continuous changes in the scene luminance (e.g. due to a flashing light), or compression artifacts.
A motion threshold can be set manually in the Motion Detection panel. Each video frame, whether in full or narrowed to the ROI, is assigned a quantified motion value from 0 to 100 in the panel. A value of 0 represents no movement, while a value of 100 indicates motion across the entire frame. For example, a frame with a motion value of 0.04, as displayed in the Motion Detection panel, would contain minimal movement.
To adjust the threshold, simply drag the slider in the Motion Detection panel. Any detected motion below the chosen threshold will not be flagged in the motion plot or appear in the bookmarks. It’s worth noting, however, that adjusting the slider does not affect the red overlay display on the video.
In the Motion Plot, the threshold is represented visually by an orange line; only motion events exceeding this line will be highlighted in the top bar. This approach provides a streamlined view of the most relevant motion events for efficient forensic analysis.
So, summarizing, you can combine the ROI to filter motion events in space and the threshold to filter events by their magnitude: this gives you a lot of control to identify relevant events.
In the image above, perhaps consider two scenarios.
You have a whole night of footage, and you need to determine when an upstairs light turns on/off.
You have several days’ worth of footage, and you need to identify every time someone enters/leaves through a front door.
Using the ROI, you can run through the footage and quickly get the answers you need.
Conclusion
In this article, we’ve reviewed all the features of Amped Replay’s Motion Detection tool. With the ever-increasing amount of surveillance footage produced, speeding up the analysis is essential for investigators so they can focus on what really matters. If you have any questions about Amped Replay or this specific feature, don’t hesitate to contact Amped Software to learn more.
