Friday, August 21, 2026

Fixing corrupted mp4 files

I went to grab some files from my Rexing dashcam, but they wouldn't play. It seems there's an issue when the dashcam powers off with the engine. So I went down the rabbit hole to see if I could recover them. Of course, most options didn't work, and some seem to be scams.

These are the files I'm working with

MediaInfo details

454295552B S2026_0817_153842_007.MP4 - simple repair

319029248B S2026_0817_165138_001.MP4 - hard repair

565644598B S2026_0817_174309_001.MP4 - reference file

Kind of worked

Clever Online Video Repair recovered the more difficult S2026_0817_165138_001.MP4 file. It recovered 314244981 bytes for $6.72, including tax. On S2026_0817_153842_007.MP4, it recovered 448,217,987 bytes for $7.35, including tax. You should be able to sign up for better pricing, but it requires sending you emails, which it seems unable to do. The video looks good, and they appear to be full-length, but the audio was not good at all. Fortunately, I did not want the audio anyway. The previews looked good, but I waited to look at the others before paying for the repaired files since I really wanted an app.

Repaired S2026_0817_153842_007.MP4

Repaired S2026_0817_165138_001.MP4

EaseUS Fixo Video Repair app appears to work from the previews, but you need to sign up for $49.95/year to get the actual files. But if you go through the app, the price is higher; then it offers 40% off if you decline the first offer. That felt scammy to me, so I didn't pay them. The interface looks more polished than Stellar's, at least.

Did not work at all

Paid

Stellar Repair for Video online chugs for a while and shows you a thumbnail that looks like it works, but then wants you to sign up for a $9.99/month subscription to get the video, or optionally buy and download the software for $ 49.99. Since it was the top-rated option I found, I bought the app that claims to fix photos as well for $59,99. It did not work on either file. A total waste of money, and buying the software doesn't get you off the "free" level online. The refund request link says it sends a code, but it didn't work for me. The interface is pretty sad too. You can't even resize the window. I'll be watching PayPal for any more charges.

Free

Untrunc free app. Did not work for either file. Its GUI is about as primitive as Stellar's. It is resizable but closes as soon as it is done so you can not see the output.

The end of the output from the command line is:

Info: Found 7980 packets ( mp4a: 4509 avc1: 3471 avc1-keyframes: 289 )

Info: Duration of mp4a: 2min 24s 288ms (144288 ms)

Info: Duration of avc1: 2min 24s 625ms (144625 ms)

Warning: Unknown sequences: 11

Warning: Bytes NOT matched: 3.26MiB (0.7517%)

Warning: Skipped atoms in mdat: 290 -> 5.66MiB total

Info: saving F:\CARDV\Movie\S2026_0817_153842_007.MP4_fixed-s1-k.MP4

src/atom.cpp:479: void BufferedAtom::write(FileWrite&, bool): Assertion `end - start == new_length' failed.

VLC Media Player: Rename mp4 to avi. Open VLC Media Player, go to Tools > Preferences > Input/Codecs, set "Damaged or incomplete AVI file" to Always fix, then try playing your MP4 (or rename it to .avi). You can also use VLC’s Convert / Save tool to re-mux the file. Neither way worked.

FFmpeg: Run the command ffmpeg -i corrupted.mp4 -c copy repaired.mp4 in your terminal or command prompt to quickly re-wrap broken headers without re-encoding. Choked when opening the files. Output:

[mov,mp4,m4a,3gp,3g2,mj2 @ 000001b2ed85adc0] moov atom not found

[in#0 @ 000001b2ed876c40] Error opening input: Invalid data found when processing input

Error opening input file S2026_0817_153842_007.MP4.

Error opening input files: Invalid data found when processing input

Tuesday, August 18, 2026

What system should I get bullet points

Not sure why people do not search the group before posting questions like this. You mainly get loads of people trying to sell you their stuff and others suggesting junk. You can hire someone to set it all up, but odds are good you will end up with their standard setup, not something tuned to your needs. What you need to know:

Avoid WiFi. When you are trying to ID or detect with AI, you want every frame. You will not even notice most frame drops until you are trying to get the one you need that is not there.

Cloud is for backups only. Record to an NVR, and it does not hurt to record to an SD card in the camera as well. My personal favorite is a PC running Blue Iris, since it is dead simple to learn, talks to about everything, and costs only $100. If you are tech-savvy, there is Frigate, which has similar features and is free but has a lot more of a learning curve to set up

Record 24/7. Anything less will miss something.

Filter with AI. You want AI to do the work of looking at all the motion triggering your cams, or you will come to ignore them. Note there will always be a trade-off between always notifying you, even when it is nothing, and missing something significant. You will need to tune this over time. Another reason to do it yourself.

Each camera needs to match its target. Look at where you want to monitor, then where you can put a camera. From there, you have distance and view angle. A site like https://www.jvsg.com/calculators/cctv-lens-calculator/ will help sort the specs you need to get the detail you are looking for. Prefer cameras that support ONVIF but require at least RSTP support.

Lighting: lots of models claim color night vision these days, but get there by slowing the frame rate down to a point where moving objects are blurred or even ghosted. Or they get there with white lights in the camera. Look for videos with moving things in them. Stills are useless. That said, the best are ColorVus, which will give you cloudy-day-looking video for a target area where you can make out your hand in front of your face. In darker areas, you will want to go with IR for night vision. It is much better to use external lighting to get the level you need, since lights attract bugs. Cameras that turn on white lights when motion is detected are a plus.

Networking: As with any IoT, you should isolate them from your main network and the internet. Either have 2 Ethernet cards in the PC, one for the camera network and one for your main network, or look up VLANs. Do not get distracted by NDAA-compliant claims. No camera should be exposed to the internet.

Thursday, August 13, 2026

Flock facts

For the record and to respond to the many uninformed commenters and posters:

The cameras are leased, not bought. $2500 to $3000 per camera per year.

The salary of one patrol cop, not including benefits, would cover the cost of 18 to 43 Flock LPR cameras per year, depending mainly on the cop's salary. Possibly more for a large number of cameras on a volume-discounted contract. The LPR models most think of as Flock cameras are pretty cheap, low-end gear that even HOAs can afford, which is why they are popping up like weeds. Flock offers better cameras, software that can connect to your cameras, and even drones.

The early termination or removal fee for a Flock Safety License Plate Reader (LPR) camera is typically $500 per camera.

Flock offers a zero-replacement plan for stolen and vandalized cameras.

The primary use for these cameras is to flag when a suspect plate is seen, not to query it. Querying the DB would mainly be to see which plates left a crime area for leads, like with any other security camera. Any query for a plate's location history is extremely suspect without a warrant tied to a case. And audit logs and role-based logins are part of the system.

The LPR cameras have a 15-foot-wide field of view at 65 feet, and are therefore only useful for monitoring narrow targets such as a lane of road, a doorway, or a path. 65 feet is the advertised maximum useful distance.

The Flock poles are breakaway models, so if you see ones that aren't, they aren't official. However, they are breakaway in that the poles are thin-walled aluminum, so it might not be obvious. The poles are the worst way to mount the cameras because they can be taken down so easily. Mounting on traffic light or street light poles is much better if available. Odds are, if the camera is on one of the thin 12-foot poles, it is either owned by someone without rights to city assets or someone fairly clueless about security. You often see posts where the town doesn't claim ownership of any Flock cameras, yet many Flock cameras are on these cheap poles in those towns.

The cameras have 2GB of soldered RAM. Not the 64GB of DDR5 RAM people are posting about.

Flock is just one maker of these kinds of cameras, and their LPR model is on the bottom end.

What is the maximum penalty for damaging a Flock camera in Texas? Yes, there could be years of jail time.

No, Flock Safety cameras do not have built-in radar or sensors to measure or detect vehicle speed. In theory, a camera can estimate your speed by tracking how far you move between frames, but that would require a high frame rate and precise installation at an exact angle to the road. Even then, the accuracy would be low. A sonic or laser sensor could be added, but radar would be more likely, since it is a simple add-on we already see in school zones and similar areas.

Flock Safety just cut its data retention from 30 days to 7 and added mandatory audit tools. The ACLU reviewed the changes today and called them "a thinly veiled PR attempt" with "largely hollow security promises."

Is this the start of Pre-Crime as in Minority Report?

The area monitoring cameras, not the LPR models, monitor for suspicious behavior, including crowds gathering. The LPR cameras have too narrow a view to be of any real use for this other than possibly on a trail watching for people that have fallen or objects left behind. Or someone waving their arms in a way that indicates they are trying to get someone's attention. One of the "LPR models" on a trail is most likely there in the mode of a security camera to provide video of anyone following someone who has been attacked or just to discourage attacks. Realistically, to be a real deterrent, there would need to be no blind spots on the trail.

The real point of contention in real life and, to an extent, in that movie, was that cops arrested people with insufficient evidence. In the movie, you will notice the minority reports were ignored. That was the system's flaw. Hence the title of the story. Same with the cameras. A confidence percentage is part of the data that is ignored. And it is never 100%. Deploying cops when a camera reports suspicious behavior is no different than deploying them when someone calls in. They have to check it out, but they also have to keep in mind it might be nothing, so they don't arrive guns blazing. With all the swatting calls these days, the cameras might be more believable than person reports, but that day is still a good while off. One report shows 40% of stops in one town were in error. In a recent incident circulating, a family was surrounded in a parking lot and removed from their SUV at gunpoint because their plate number was reported stolen. The plate was for an out-of-state motorcycle. Clearly, the cops did not confirm any details before approaching the vehicle, which should be standard procedure. Especially with kids in the target vehicle. Another similar case: a vehicle was stopped on a partial plate match of just 2 digits. There are also cases of people detained on facial IDs despite it being easily provable they were nowhere near the crime area. In a Florida case, 300 miles away. Lawsuits against those police departments are, of course, pending.

Oddly, the movie Mercy (2026) is probably a better fit for where this tech is headed. Yet to see a post suggesting that though.

Other info:

Flock's 6 Myths About License Plate Readers and Security Systems

Dave's Garage: The Controversial Flock Cameras Tracking Every Car — Full Breakdown

Flock's FAQ

Deer Park TX PD Transparency Portal An example of details of the volume and types of data collected and what it was used for.

Flock 2020 patent for Method and system for capturing and storing significant surveillance images. This seems like what most NVRs have been doing long before 2020, so I'm not sure how they got a patent for it.

You can look into tools like Have I Been Flocked? to see public tracking of known ALPR audit logs and searches.

deflock.org is a crowdsourced DB of Flock cameras.

In Chatrie v. United States, decided on June 29, 2026, the U.S. Supreme Court ruled 6–3 that police conduct a Fourth Amendment search when they acquire a user's cell-phone location data through a reverse-location "geofence" warrant, even if a third party holds the data. This is often cited as making Flock cameras illegal, or at least requiring a warrant to search the DB for plates in the area where a crime was committed. But it is hard to say right now,

Texas Privacy Coalition has links mainly to anti-Flock posts.

Snoopes: Real shirt that confuses Flock cameras, other surveillance tech hasn't been sold for years . It doesn't mention that LPR cameras are generally aimed at road lanes, so you would need to walk in the road, and the plates on the shirt would quickly get added to an ignore list. See "whitelist" in the Wired article below. On a trail, an LPR hit should automatically trigger an alert if it isn't whitelisted, like known park vehicles or these shirts.

Wired: Flock Has a Powerful New AI Tool for Police. We Got Its Code : Shows the queries the system supports. Note: the article is heavily slanted, implying it reveals secrets they uncovered rather than features promoted on Flock's site. The stock queries are interesting to look at. They all seem to have legit crime-solving uses. Also, many of them create lists from police data, not Flock camera data.

Video: Removing Tracking From Your Car, Disabling Cellular Radios & And Other Trackers Talks about cellular radios in cars that can be tracked as well. Doesn't tell you how to remove but this one does How Smart Owners Are Pulling Trackers From Ford, Toyota, and RAM (The Module Delete) You should note Flock, and probably others, are starting to log the cellular IDs with plates. So a vehicle that ought to have a radio but does not will probably trigger an alert at some point.

Coupling LPR with signals data: SignalTrace has been rumored to be linked or even integrated into newer LPR models. Lexipol has a more balanced take.

Texas investigating Flock Safety over claims it operated without a license

Friday, May 22, 2026

Q: I am leaning toward ReoLink's 12K cameras

A lot depends on your lighting. Reolinks, as a rule, are not great at night vision, mainly due to limited config options. I do like their POE TrackMix model though. Two lenses, so you get a zoomed and a wide stream, plus the wide stream helps with tracking in the zoomed view. Better IR night vision than many of their models too.

Do not bother with high MP cameras. Even 4K is often overkill, but 4MP generally looks much better than 3MP due to the sensors involved. Cameras are focused on a targeted range. Adding pixels generally does not let you zoom in much farther, though they might give you a wider view at the target range. That target focus range is not always specified in the specs for fixed-focus models, but it can often be inferred from the DORI numbers. Autofocus can help here, but be aware that a bug or bird might cause the camera to refocus, and it might not focus where you want it to. So, it's best to get it to focus on the target area and lock it there unless it is a tracking model. Never try to cover more area than what your cameras can see without moving.

For choosing models, the same rules always apply. Sort the distances to the target for each camera and decide how much detail you want to see. Match that to the DORI numbers in specs. Then factor in the width of the view angle to cover that target area. Lastly, factor in the lighting needed. Especially for color night vision, you want to compare min lux values and see real video. Many push the specs by reducing the exposures down to the point of ghosting moving objects. Or claim with internal white light on. IR night vision is not as bad. In both cases, you generally want to run without the internal lights. Accent lights for color, or IR floods for IR, are the way to go to keep bugs from visiting your cameras. Oh, and they should support RSTP or, preferably, ONVIF, which most of them do. Ignore the NDAA compliance BS. Do not assume ANY camera is safe, and block them from accessing the internet.

For NVR, go with a PC and Blue Iris or Frigate. Blue Iris has an easier interface and has an extendable AI, but is Windows-only and $100 for up to 128 cameras. Frigate is free and runs on Linux, but can be challenging to set up for non-IT types. Both work with cameras that support RSTP or ONVIF.

Tuesday, April 21, 2026

Updated copy/paste for what camera or system should I get.

While I love Unifi network gear, their cameras seem pricey and feature-limited.

For a doorbell, Reolink worked best for me (over Ring and Hikvision, which make most of the POE ones).

For other cameras, sort the specs to see the details you want in footage. Look for DORI numbers in the specs and ONVIF, or at least RTSP. Note that, as a rule, Reolinks do not have great night vision and offer limited image control. For color night vision, it gets tough. Many cameras, like Tapo, for example, crank down frame rates to brighten the image, which causes ghosting. Best is still Hikvision ColorVus, which will produce cloudy, day-like video if you can make out your hand in front of your face in the target area. Otherwise, if you can't see to walk without a flashlight or the moon, you want a camera that has IR for night vision.

Note: Avoid turning on in-camera lighting as it draws bugs and will white out close objects. Used IR floods or accent light as needed to add illumination.

Tracking is helpful, but it's best as an add-on. You want total area coverage recording 24/7 because no detection is perfect. Note: Reolink has a nice dual-lens tracking camera, but avoid the WiFi model.

NVR-wise, Frigate is free, and a lot of people seem to like it despite the setup pain. I tend to suggest Blue Iris for easy setup while still talking to most cameras and third-party integrations. I'd avoid all standalone NVRs. Especially Reolink's, which sets unknown passwords on the cameras, preventing direct access.

And of course, block all the cameras from the internet and the rest of your network. Preferably, only talking to the NVR PC.

Wednesday, August 6, 2025

Who is flying overhead?

Especially since the July 4th flooding. There have been a lot of helicopters in the sky over my neighborhood and a lot of posts on NextDoor and Facebook asking, "Who is up there?" So here is a quick guide to finding out.

Planes and Helicopters

Start off by going to https://www.flightradar24.com/2025-08-05/22:44/20x/30.58,-97.96/14.

Where:

2025-08-05 is the date you saw the aircraft in year-month-day format.

22:44 is the time you saw it in 24-hour format and the UTC timezone (add 12 hours for PM times and another 5 hours for Central Daylight Time, summer, or 6 hours during Central Standard Time, winter)

20x is the zoom. 20x is a good starting point.

30.58,-97.96 is the GPS to center the map on. (This is my neighborhood.)

14 is the zoom level. 14 is a good start for things overhead. If you are looking closer to the horizon, you will probably want to go to 11, which shows a wider area.

Click the play button to see the animation. It then turns into the pause button.

Once you see a likely aircraft, hit pause. Click on the craft's icon for info about it.

You can then click on "Route" to view the entire flight path.

Entering the tail number (R20045 in this example) in the search probably will return nothing because it is a military helicopter. So I'm using N3010X instead to show what a non-military aircraft would yield.

Clicking on that results gets you the history of its recent flights and basic info.

Sometimes adsbexchange.com has better info, though it is not as user-friendly. The equivalent URL for it would be.

https://globe.adsbexchange.com/?replay=2025-08-05-22:46&lat=30.430&lon=-97.933&zoom=11.0

Clicking on the military helicopter provides a bit more information.

Clicking on FULL DETAILS provides more information.

Drones

Drones are supposed to be using transponders now, too, but I would not count on them. You should see them on FlightAware, but I haven't seen any listed anytime I've looked. There are also apps that seem not to work well, according to reviews, and insanely expensive devices from Air Sentinel and DroneTag. Dedrone will not even disclose the price without contacting a salesperson, which usually means OMG pricing. Going DIY with an SDR dongle seems reasonable unless you are having issues with drones flying without IDs. Plus, that could give you a logging option that many of the others lack.

Monday, October 31, 2022

Another attempt at a quick paste response to "I want something with good night vision"

Best color night vision is Hikvision's ColorVu G2s. If you want IR night vision there are more options. I prefer Dahua or JideTech myself but many use the same larger sensors. For IR get the largest sensor you can. 1/1.8 (like Dahua IPC-HFW4631H-ZSA) of better if you can. Forget about stated LED ranges since you will want to turn those off anyway unless you like attaching bugs to your cams. Use seperate lighting like accent lights or IR floods mounted away from the cams.

Plus of course match the specs to the distance and view angle you need. But keep in mind most fixed focus cams are focused for 10-30 feet, depending on model, so for over 50 feet plan on something you can focus for the target range either motorized (usually zoom models only) or manually. Don't assume fixed model are easily adjusted as most are glued in place these days.

What ever you get make sure they support ONVIF or at least RTSP so you can mix brands as needed.

To dig deeper look at https://securitycam101.rmrr42.com/2020/05/getting-cam-specs-easy-way.html for info about how to use an online GUI tool tha lets you set up virtual cams on your site to get an idea of what diff models will be able to see.

Wednesday, September 28, 2022

DeepStack vs Sense AI and improving recognition.

First they both use:

YOLO models to compare found objects to.
The Sense AI API is a superset of the DeepStack API so something written for DeepStack should work with Sense AI too.
They use the same coding languages.

The main diffs between DeepStack and Sense AI are:

Sense AI is considered to be the more stable and supported (going forward) version. Though as of this writing you can probably find a lot more about working with DeepStack.
Sense AI will support more types of recognition and secondary operations like OCR.
But Sense AI is basically beta at this time so the real advantages are more subjective.

As explained in my DeepStack posts a lot of your success depends on:

Mode engine is run at. Medium is default. High will yield better results but at a cost of higher load. The reverse is true of Low.
Mode model was created for. The model ought to be made for the same or higher mode than the engine will be run at to see improved performance.
Trigger image might not be the best for recognition. You will want to tweak number and frequency of images sent to the AI when triggered to get better confidence in what is detected.
Same goes for minimum confidence percentages. Too low and you will get lots of wrong detections. Too high and you will miss things. Best to start low and see what object you want to know about are detected as what and at what confidence.
How close objects used in training match the ones in images used in detection. This is the biggie and why so many look to create their own custom models. Both for object types not in the default model and to more closely match expected targets in lighting and coloring. You can find tools for simplifying making your own custom models in my deepstack repo.

Friday, December 17, 2021

DeepStack training

My test set

My current test set can be downloaded from here.

Hardware

This workstation is set up for video editing so it is pretty fast. It was in it:

3.50 gigahertz AMD Ryzen 9 3950X 16-Core
64 GB of RAM
Samsung SSD 970 EVO Plus 1TB [Hard drive]
NVIDIA GeForce RTX 3080

Dataset prep

If you are looking to prep your own dataset you might want to look at the utils I have here.

You may want to look at Why Image Preprocessing and Augmentation Matter as well.

Method Compare

I used the original 91 pics with 194 labeled objects of 4 types of my test set:

Colab 2.973 hours
Jupyter hybrid method 18 minutes
local 0.299 hours (just under 18 minutes)

The models created appear to be the same. The latest model with the latest data set can be downloaded from here.

Local

Start here. Took me a while to sort what I was doing wrong as the instructions can be a bit misleading and python being at least 20 something on the list of languages I use the most. I finally figured it out while trying to debug the "local Colab" / jupyter method.

Be sure to install CUDA and CUDNN not just one. Remember the versions.

PyTorch

Instead of the suggested something like

pip3 install torch==1.10.0+cu113 torchvision==0.11.1+cu113 torchaudio===0.10.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html

Instead try

pip3 install torch torchvision==0.10.1+cu111 torchaudio torchtext -f https://download.pytorch.org/whl/torch_stable.html

The +cu111 bit seems to align to Cuda 11.1. Leaving off the other == bits lets it sort versions it needs itself.

Download training data to deepstack-trainer../deepstack/train

So assuming your workspace is C:\DeepStackWS

deepstack-trainer would cloned in C:\DeepStackWS\deepstack-trainer

and your training data would be in C:\DeepStackWS\deepstack\train

Then cd deepstack-trainer and run

python3 train.py --dataset-path --exist-ok "../deepstack"

Running this way not only avoids the notebook framework making automation easier but puts more progress info to the screen to see how things are going.

Options from train.py (with my notes)

--model, default='yolov5m', help='yolo model to use' (best option most cases)
--classes,default='', help='model.yaml path'
--dataset-path, required=True, help='path to dataset folder' (Should contain at least a train folder. May also contain test and validate folders. If folders exist they should contain some image files and copy of classes.txt or programs will crash.)
--hyp, default='data/hyp.scratch.yaml', help='hyperparameters path'
--epochs, default=300
--batch-size, default=16, help='total batch size for all GPUs' (>16 did not work for me)
--img-size, default=[640, 640], help='[train, test] image sizes'
--rect, help='rectangular training'
--resume, default=False, help='resume most recent training'
--nosave, help='only save final checkpoint'
--notest, help='only test final epoch'
--noautoanchor, help='disable autoanchor check'
--evolve, help='evolve hyperparameters'
--bucket, default='', help='gsutil bucket'
--cache-images, help='cache images for faster training'
--image-weights, help='use weighted image selection for training'
--device, default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu'
--multi-scale, help='vary img-size +/- 50%%'
--single-cls, help='train as single-class dataset'
--adam, help='use torch.optim.Adam() optimizer'
--sync-bn, help='use SyncBatchNorm, only available in DDP mode'
--local_rank, default=-1, help='DDP parameter, do not modify'
--log-imgs, default=16, help='number of images for W&B logging, max 100'
--workers, default=8, help='maximum number of data loader workers' (>8 did not work for me)
--project, default='name folder dataset folders are in', help='save to project/name'
--name, default='exp', help='save to project/name'
--exist-ok, help='overwrite if not first run instead of making new folder' (runTrain.bat assumes this to copy model to DeepStack instance)

Note increasing --batch-size or --workers caused and out of resources exception for me despite having plenty of unused RAM, GPU, CPU and HD. Though there maybe some missed option as Python only appears to be using 1 GB of RAM.

Gotchas

C:/w/b/windows/pytorch/aten/src/ATen/native/cuda/IndexKernel.cu:97: block: [0,0,0], thread: [95,0,0] Assertion `index >= -sizes[i] && index < sizes[i] && "index out of bounds"` failed.

This is caused by a bad mapping file with a class ID larger than the number of classes in the train\classes.txt file.

Note there is a similar problem using the LabelImg program. It will crash if it reads a map file with an ID larger than the number of classes in labelImg\data\predefined_classes.txt

Add debug

If you are seeing another kind of crash or odd behaviour with train.py try changing the set_logging method in deepstack-trainer/utils/general.py to look like this

def set_logging(rank=-1):

# logging.basicConfig(

# format="%(message)s",

# level=logging.INFO if rank in [-1, 0] else logging.WARN)

logging.basicConfig(

format="%(message)s",

level=logging.DEBUG)

Colab Method

The DeepStack docs mention doing this but they kind of assume you know what you are doing with Colab.

Open My edited copy of their notebook for cloud training. Notebook in this instance means project file not notes on how to do something which might be your first guess looking at it.

Click on copy to drive to get a copy you can edit.

I want to have my data and output persist between runs so I'm mounting my Google Drive for storage.

Go to / click in the mount drive section and press Shift-Enter to run it

Go to / click in the clone trainer section and press Shift-Enter to run it

Section 1

This takes a while. (Mine said 4 minutes.)

Note it loads old versions of packages so I was left with this error/result:

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. torchtext 0.11.0 requires torch==1.10.0, but you have torch 1.7.0+cu110 which is incompatible. Successfully installed dataclasses-0.6 torch-1.7.0+cu110 torchaudio-0.7.0 torchvision-0.8.1+cu110

While this running you can work on preparing your data. See Preparing Your Dataset. If you just want to try it before creating your own data set you can find mine here. It is the one I'll building on to detect the wildlife types I have here.

Finally run the Add Your Dataset section which will take a fair bit of time. My set of 91 images took ~30 seconds per run and the default is 300 runs.

Note if you do not scroll the output often enough you will get a popup asking if you are still there. At which point I think it pauses your processing till you respond. So do not leave it to run overnight and expect to be done in the morning.

I finally got this result

Epoch gpu_mem box obj cls total targets img_size

299/299 9.26G 0.01827 0.0114 0.001359 0.03103 59 640: 100% 6/6 [00:30<00:00, 5.07s/it]

Class Images Targets P R mAP@.5 mAP@.5:.95: 100% 1/1 [00:00<00:00, 2.16it/s]

all 11 0 0 0 0 0

/usr/local/lib/python3.7/dist-packages/seaborn/matrix.py:198: RuntimeWarning: All-NaN slice encountered

vmin = np.nanmin(calc_data)

/usr/local/lib/python3.7/dist-packages/seaborn/matrix.py:203: RuntimeWarning: All-NaN slice encountered

vmax = np.nanmax(calc_data)

Exception in thread Thread-10:

Traceback (most recent call last):

File "/usr/lib/python3.7/threading.py", line 926, in _bootstrap_inner

self.run()

File "/usr/lib/python3.7/threading.py", line 870, in run

self._target(*self._args, **self._kwargs)

File "/content/deepstack-trainer/utils/plots.py", line 122, in plot_images

colors = color_list() # list of colors

File "/content/deepstack-trainer/utils/plots.py", line 32, in color_list

return [hex2rgb(h) for h in plt.rcParams['axes.prop_cycle'].by_key()['color']]

File "/content/deepstack-trainer/utils/plots.py", line 32, in <listcomp>

return [hex2rgb(h) for h in plt.rcParams['axes.prop_cycle'].by_key()['color']]

File "/content/deepstack-trainer/utils/plots.py", line 30, in hex2rgb

return tuple(int(h[1 + i:1 + i + 2], 16) for i in (0, 2, 4))

File "/content/deepstack-trainer/utils/plots.py", line 30, in <genexpr>

return tuple(int(h[1 + i:1 + i + 2], 16) for i in (0, 2, 4))

TypeError: int() can't convert non-string with explicit base

Exception in thread Thread-11: