17-year-old Kendrick Johnson’s dead body was found rolled up in a gym mat at Lowndes County High School in Georgia. The family and investigators are still trying to figure out what led to his death. Everything is in dispute, from the cause of death to how many people were in the gym when he died. The case, which has received widespread media coverage, has now zeroed in on the school’s surveillance footage.
Why cant the video investigation be done onsite directly using the playback system and then exporting what is required? Is there an issue for law enforcement to be onsite and sit in a room doing forensics? Other than that, just copy the whole drive and hand it over.
Certainly, some can be done on-site but going through all of that video would will take a long time and multiple iterations as the investigation unfolds. Law enforcement can certainly spend some time sitting in a roombut spending hours per day across a number of weeks is typically logistically challenging.
It's more complicated than just copying the 'whole drive'. With 32 cameras, there are probably a number of drives, depending on who they archive, video for the same days could be spread across them, etc. This is why typically people use the recorder's built-in exporting function.
Why would it take a long time to go through the video? I assume it's time stamped and everyone knows the approximate times the incident happened so a review shouldn't take very long.
The police should have insisted on direct access to the recordings with at least some expert there to check for corrupted files. If data problems were subsequently found, the evidence could have been preserved by either protecting the video from deletion or stopping recording.
A court order could have been obtained fairly easily. It sounds like both the school and the police have some 'splaining to do.
In a city of 50,000 the police department would have no less than 60 personnel (using the staff-per-thousand population minimum expectation) and the fact it was a homicide would have been instantly obvious to the first detective responding. The fact there was video available would have prompted even a newbie detective to immediately review the video in person (or delegate that duty to another commissioned officer) with whomever on school staff was trained to operate the VMS. It would be the detective's discretion to take the video server into evidence or have the entire HD mirrored. This case has so much potential for both the school district and police department to look like rank amateurs at best. The GBI is a well-trained instant resource as the state's supreme law enforcement agency to assist any local agency that asks. Apparently in this case the PD didn't follow basic crime scene tenants so who knows where the investigation will lead.
Watching the video, it appears perhaps it was set up to record on motion and the motion detection settings we set very conservatively or incorrectly. The fields of view aren't very useful in this case (perhaps set to detect intruders than surveil the gym), and there is no excuse for a camera being that badly out of focus.
Unfortunately, its far too typical for a public school installation, low bid got the job, and the system is probably serviced and repaired by a the school maintenance staff that aren't familiar with video surveillance systems. It sad as a properly designed, maintained and working system might have been useful in prosecution or bringing closure to the family.
Is there any information on the type of system and cameras used or the contract for the installation?
John, I agree about the motion detection incorrect settings. If it is that, which would not surprise either or us, it's interesting to see how the public immediately assumes it was something nefarious when it was really incompetence.
I was actually looking for some of those documents last week. I called the school system when I was writing the postand they were all out forbreak but a receptionist said all of that should be online. Unfortunatelytheir online system is pretty heinous and not searchable. I haven't given up though. It would be interesting to know the exact equipment.
This brings up the question of Auto Focus features on many newer cameras. Is it best practice to set the focus settings on the camera web configuration to manual and lock them in at what appears best at time of install? Or allow the camera to refocus after Day-Night switching and, if available (as it is on Pelco cams) auto focus on temperature changes? We have installed some cameras with this abbility (in a school I might add) and we occassionally get complaints that the image is fuzzy. By the time we have gotten out to check it out, the focus is good again since it is refocusing twice a day due to the day-night switch. I wonder if that could play a factor in a case like this where the camera may not have acheived good focus on a refocus attempt. Now I have something else to lay awake and worry about at night! What are other integrators doing as best practice with the Auto Re-focus capabilities of newer cameras?
I would assume this is an analog 32 channel system. I think autofocus abilities are a bit of a reach. I will attest to the fact that I turn Pelco auto focus off because of night and day out of focus issues. I can not even count how many times I have come across dx8100 systems installed with defaults. That is quarter resolution with record on motion only. Sensitivity threshold at 50 (when it should be much less). Shame on the manufacturers for at least not setting better defaults with a normal record of at least 2fps. In gyms, motion should be used with very long prerecord and post record settings. We set up all exterior, gym, and cafeteria cameras nowadays to free run. My money on this one that the low bid skimped on storage to a 250 or 500 gb drive. Still, the taxpayers paid for a system that was poorly setup (recording gaps) and was poorly maintained. If I was the parent, you would think I could have a lawsuit against the police, school, and the integrator that installed the system.
I was involved on an incident that involved loss of life on one of our customer's site several years ago and the homicide detectives immediately seized the DVR recording any camera covering the area where the incident occurred as evidence. Maybe there are differences accross states on how to legally do this.
The school gave an exclusive interview to the local paper and said the time's were not synced up on each server because it had never pulled enough video to realize they weren't synced, which would explain the reason the times don't match.
As for missing footage, it said it was because it just didn't pick up all of the events. As we mentioned above, the cameras are set to record on motion. The school says often there isn't enough motion to trigger the cameras or if a person doesn't take up enough of the FoV it won't register as motion. When the video stops recording while motion is going on, the school says the subject may have moved too far away to take up enough pixels and the camera doesn't register the movement.
On the motion detection configuration, typically manufacturers default to overly sensitive so that the system does not miss small movements but suffers from more nuisance motion events. This is, of course, to ensure situations like this do not occur. To that end, I wonder if this system had poor defaults or if the admin/integrator turned the sensitivity down to reduce storage consumption. Either way, this is a classic example of the fears people have depending on motion detection. See our Surveillance Recording Mode Statistics
I don't want to say who, but I noticed on a major DVR manufacturer we deal with they would default to max framerate, 2 or 1 CIF resolution at medium quality, motion only recording.
It seems their thinking was to make the video fluid with high framerate like you see on TV and in movies, but hook that customer with high potential retention periods, thus the motion based recording at reduced picture size.
I think the industry, mostly manufacturers, have painted themselves in a corner with their emphasis on high framerate capabilities at the same time trying to appease customers desire for long retention times, when the majority of time low framerates are all that are needed and it's the quality of the still picture (when you pause the video at climax of the incident) that customers really care about.
There are a number of issues here And I can only comment from my own experience and the instructions I give to investigators.
In major incidents, as a video investigator, I want to visit the scene, view the DVR in situ, document all the details, and conduct the recovery of evidence myself. I have stated in the thread regarding police seizure that in certain circumstances I do remove the entire DVR, but in most circumstances, I replace with another.
I regularly have to deal with motion activated recordings and am asked the question of why was something not recorded. We conduct motion testing within the environment to identify when the recordings 'kick in'.
I can understand why motion detection is used but in most serious cases, it causes more problems and questions.
Some of the best systems I see only use motion detection when the premises is not being used And then it has a good buffer so it's always recording a small period of time so you always get the start of an incident.
Reading the reader comments on CNN is always entertaining. This is from one of the follow up articles on the Kendrick Johnson's death:
I'll preface this by saying I think the locals deserve all of the scrutiny they are getting, for the sloppy way in which they handled the case. I won't speculate on motive, but it seems pretty clear at this point that the locals did not take this case seriously at all, and are now in CYA mode.
I will say as someone who runs motion capture IP (network) cameras at home, I am not surprised at all that the footage looks so "sketchy".
A few fun facts I learned from wrangling with IP cameras:
1. Most IP cams are cheap, chinese-made junk and are prone to malfunction.
2. Footage takes up a LOT of space, even with good compression. That is one reason why most surveillance footage looks so crappy- lowering framerate, lowering resolution, switching to black/white are all space-saving techniques (also, see point #1). It is common to record at 4-5 frames per second- just enough to get some frames to identify people.
3. jumpy/missing footage is normal if these cameras were set for motion detection. Usually it takes several frames of motion for an IP camera to "kick in" and start recording, and you will lose those first few frames unless you have a buffer system set up in the camera or on the capture server (which takes more resources). If your cameras are capturing at 4-5 frames per second with no buffering, you will lose the first 1-2 seconds of any motion event you record (which also makes the things damned useless for someone running by the camera).
4. Lastly, because these seem to be network cameras, you can get dropped frames from lag or network congestion (or from using crappy cameras).
The school or sheriffs certainly could have tampered with footage, but the footage we do see is pretty standard fare for a large array of network cameras.
Exactly my point Carlton. A lot of times unfortunately, the folks handling things like these are no more aware of the capabilites of a modern IP video system than the guy who posted the comment I published above because that is how the cameras he installed himself at home work.
There should have been an immediate seizing of all recorded material at the time of the death, not 11 months later like John mentioned. Even if there was no suspicion of foulplay at the time, just for the sake of telling the family what happened to their kid, the video footage should have been investigated or at least protected; hopefully they did that. This is what our local PD did on the case I described on my comment a couple of weeks back.