Run a motion analysis, select a region of interest or set a motion threshold. Find out how easy it is to find suspect movement in Amped Replay!
Hello folks and welcome to a dedicated article on Motion Detection, the newest feature of Amped Replay. When we first came up with the idea of developing this software, we wanted to first and foremost assist law enforcement officers with speeding up their video investigations. We appreciate how long and tedious the process of watching endless CCTV clips can be, especially when you are not even sure to find something of interest in the first instance. It is a job not many want to do but it has to be done and is so important, especially for big-scale investigations.
Amped Replay already has several intuitive and easy-to-use features to aid investigators deal with video evidence, even if they are not computer and/or video experts. But we have now implemented that missing piece of the puzzle, which we believe will really make the difference! Let’s go and find out how to find movement in Replay.
The Motion Detection Panel
When you navigate to the Play tab, you will now find a dedicated Motion Detection panel in between the File Info and Case Info panels. This is where you will analyze and customize motion events in your video clip.
First of all, load any video clip into Replay, short or long that it may be. Of course, you will truly appreciate the value of this new feature with long CCTV clips. Firstly, because it is more difficult to spot isolated and short bursts of movement in long videos. Secondly, because smartphone recordings, for example, will predominantly contain movement since the camera is not static.
Starting the Motion Detection Analysis
To start the motion analysis, simply click on the first button of the Motion Detection Panel, showing as a “running man” icon. The clip will now be analyzed and a progress bar will be displayed, indicating the progress of the analysis.
When completed, you will see red blocks overlaid on the regions where motion has been detected. In the sample above, the two walking subjects (identified with blue arrows) have been correctly detected as moving. The two ladies on the pavement at the far side (highlighted by a yellow square) have not been detected, as they are actually standing still in this frame. The male at the bottom right (identified with a green arrow) is moving his right arm in this frame. This has also been detected correctly. Pretty cool, right? But how can Amped Replay do this?
Video motion detection works by comparing sequential images from your video. If there have been enough pixels that have changed between those video images, the tool decides that something has moved and highlights it as a motion event. There is no neural network involved in this decision-making. The process uses numerical comparisons of pixel patterns between consecutive frames and therefore it is forensically secure.
Once the motion analysis has been processed, you can also hide the red color overlay or bring it back at your convenience. Simply click on the same button that we previously used to run the analysis. You will also need to click on this button when re-opening the clip in Amped Replay to rerun the analysis, if you wish to detect motion again. Re-running the analysis is not necessarily needed, though, since motion values for each frame are already stored in the project file. You’ll need to re-run the analysis only if you wish to view the red overlay, or if you want to change the selected region of interest (more on this below).
Selecting a Region of Interest
With the new feature, it is also possible to specify a particular Region of Interest (or ROI) for the motion detection analysis. This can be particularly useful if you suspect or expect that a subject of interest or a vehicle has appeared in a particular area of the recorded scene at some point in the video. Maybe you have some intelligence provided by a witness that a robber has exited the crime scene from a specific footpath in the scene. A suspect vehicle may also have turned at a known intersection recorded on your CCTV footage. Or you may just wish to detect movements around a shop entrance and not around the aisles.
To narrow down the analysis and reduce the amount of false positives, we can then draw a rectangular selection of any size and position in our video. The motion detection will only take into account that specific selection. Please note, however, that you will first need to run the analysis on all pixels before you can specify a ROI. To select a ROI, click on the second button in the Motion Detection panel, showing as a dotted rectangular icon. Then, simply click and hold the mouse button on the viewer and move the mouse to drag out the rectangular selection. When satisfied with the size and position of the ROI, simply release the mouse button.
The ROI will be illustrated as a solid magenta rectangular outline on your video. You can draw the ROI as many times as you like until you are fully satisfied with the area that you have selected. However, note that you can only have one ROI at a time (for now!). If you want to clear the ROI and revert the analysis to all pixels in the video, simply click on the next button along. This button is shown as a crossed rectangular dotted icon. The current ROI’s coordinates are shown in the Motion Detection panel, and they are included when you generate the report.
Alright, so we have run the analysis and set a ROI. How do we access and flag all these motion events that we have found? We can simply navigate through the detected events by clicking on the Previous Event and Next Event buttons in the Motion Detection Panel. These are shown as left and right arrow icons, respectively. Each motion event is identified by a starting frame and an ending frame. Clicking on the arrows will let you navigate through them.
Or, even better, we can bookmark them so we will have a visual representation of where the frames containing detected motion are in our timeline. The motion detection bookmarks are magenta, so you can visually distinguish them from any conventional bookmark that you may have previously added.
And remember, whenever you add bookmarks, they are included in the Amped Replay report if you decide to generate one. The bookmarks will appear as frame numbers and also as illustrations. And of course, you can also export all the bookmarks individually as images.
Viewing the Motion Plot
This is a graphical representation of the amount of motion present in your video. It is designed so that you can quickly identify portions of the timeline with high amounts of motion. You can also easily spot other portions where there is less motion. The motion plot can be activated at any time. However, you will need to have run a motion detection analysis as previously described, before you can do so. Simply click on the Motion Plot button located at the bottom of the Replay interface. You will be able to click on this button whether you are in the Play, Enhance, Annotate, or Export tab.
You will now be able to monitor the motion plot in your timeline. Video frames containing a high amount of motion will be identifiable by high peaks in the plot. Conversely, frames containing little or no motion will have lower peaks.
The navigation through this plot is very similar, in terms of functionality, to the audio plot which we discussed in a previous blog article on audio redaction. The bottom bar shows the motion plot for the entire video, whereas the top bar shows a zoomed-in view of the plot for a specific portion of the video. You can zoom in and out of the plot by hovering over the top bar and by using the scroll wheel on your mouse. You can still scroll through all the frames in the video by dragging the play head left or right, as required.
Areas of the video with motion amount exceeding a set threshold will also be highlighted on the top bar of the motion plot, as explained in the next section.
Setting a Motion Threshold
To ensure we are flagging up the motion events that are really of interest to our investigation, we may need to set a motion threshold as an alternative to, or, in addition to setting a Region of Interest. The video may contain so many frames with a high amount of overall motion, that we may want to be more selective in what we flag up as movement.
To do this, we can manually set a motion threshold from the Motion Detection Panel. For each frame of the video, in full or narrowed down to Region of Interest (ROI), the amount of overall motion will be quantified in the Motion Detection Panel as a numerical value ranging from 0 to 100. A video frame with 0 motion will have no movement at all. Conversely, a frame with 100 motion will be made by moving pixels in its entirety. In the video below, for example, the frame selected has a motion value of 0.04, as displayed by the Motion Detection Panel, i.e. containing very little motion.
We can increase the threshold to filter out motion events detected for negligible reasons. This includes small objects rattled by a little wind or compression artifacts changing their appearance between frames. Simply drag the slider located in the Motion Detection Panel. This will ensure that events with detected motion below the set threshold will not be flagged up in the motion plot or when you add bookmarks. Note that moving this slider, however, will not change the rendering of the red color overlay on the video.
The threshold value will be visually represented in the plot by an orange line, as you can see in the illustration above. Only motion events that are above the orange line will be highlighted in the top bar of the motion plot.
Conclusions
And this brings us to the end of this article. We really hope you find Motion Detection useful and that you will be using this new tool when triaging your CCTV evidence. Feel free to let us know how you are getting on with the feature or if you are having any issues with it by joining our Discord channel. Until the next time, please stay safe and take care.
