Hey sentdex, thank you for another amazing tutorial. In India today is called 'Teacher's day', this is the day we take time to appreciate and give our respects to anyone who has helped us gain knowledge. To me you are one of them, I just want to thank you for all the amazing context that you put up. It has certainly helped me a lot.
@ThomasSMcNamara7 жыл бұрын
I am just getting into Python and really like it. Though everything you are doing is way over my head, I love it. It is inspiring and just really fun. Thanks for sharing this with us.
@pottingmix37337 жыл бұрын
Everyone starts somewhere as long as you don't give up you will be as good as anyone in python. Remember practice makes perfect :P
@sundaresann47497 жыл бұрын
Hi, i suggest to watch python3 tutorials @sentdex...even i learnt lot from it.
@ThomasSMcNamara7 жыл бұрын
Very cool! Thank you for the suggestion, I will do that!
@ThomasSMcNamara7 жыл бұрын
Thank you! I am really enjoying it.
@arjunkrishna88737 жыл бұрын
go through his machine learning with Python series kzbin.info/aero/PLQVvvaa0QuDfKTOs3Keq_kaG2P55YRn5v you could learn everything from scratch there
@wolfisraging7 жыл бұрын
You are the best programmer I have ever seen, your videos are so much precious and unique, no one can find any such materials on KZbin, but you are such a n awesome man, I really appreciate your work, I love you, and I am learning a lot from you daily. Big fan😊
@sentdex7 жыл бұрын
Thank you for the kind words :)
@Davidnguyen-ck5bv6 жыл бұрын
just start tensorflow and machine learning in less than 6 months! guy like you who is willing to give free code guide help me alot!!! Greet from Vietnam and have a nice day!!
@protodaishi62277 жыл бұрын
I watched a handful of your tutorials a while ago but then kind of got out studying for a while. Starting to dip my toes back in and I think going back through some of the same stuff is a good start. It's really cool to check back and see you are still doing cool stuff. I'm glad you are doing something you love and you are doing a huge service to people. Thanks :)
@chandrakantpattekar24657 жыл бұрын
You are the best Sentdex! This video series employs many concepts of AI and I want to say a big thanks to you for bringing all this together. Eagerly awaiting your next tutorial video.
@DisgracedCubFan7 жыл бұрын
This is the coolest thing I've ever seen.
@sonusharma-uc7qb4 жыл бұрын
These tutorials are best in every way , it motivates and provide different aspects of using things , thank you for such a wonderful series
@Mr8perezm5 жыл бұрын
I've been watching your zipline tutorials and came across this video. Very impressed, you're pushing python and general programming to the limit. Very inspiring video for newbies and kids.
@digvijaysingh63327 жыл бұрын
Thanks sentdex for spreading your knowledge. It really motivating me.
@digvijaysingh63327 жыл бұрын
Thanks for your quick response :)
@sundaresann47497 жыл бұрын
You're really helping me to develop my career...:);)
@sentdex7 жыл бұрын
Awesome to hear!
@alsdmaslm5 жыл бұрын
Same here! I'm currently preparing for a job interview that involves machine vision!
@inimk7 жыл бұрын
Man you got a good memory power to hold that knowledge.. I learn so much stuff but can't apply even 10% of it..
@maximkhanov47887 жыл бұрын
I'm so happy that you post so often! Enjoy your videos keep up the good work!!
@xuanzeng34267 жыл бұрын
Like the way you keep coming up with interesting ideas in exploring the topic with us!! Thank you
@Fiifufu7 жыл бұрын
Teaching robots to steal cars and run from police since 2017
@sundaresann47497 жыл бұрын
You're genius harry..Hats off to your work as i said you before, even kids can easily learn from you..true ;).. Thank you...::::))))
@DavidChipman7 жыл бұрын
I like the fact that the image detection system thinks your speedometer is a clock! ;)
@sentdex7 жыл бұрын
It's fairly close at least. I think the clock class could be renamed to "dial." I expect it was just trained on clocks, but it seems to detect any dial as a clock.
@DavidChipman7 жыл бұрын
That's kinda what I thought. Just thought it was funny.
@IsalaPiyarisi7 жыл бұрын
Dudeee !!! This is sooo freakin awesome please continue releasing tutorials like this
@tobiea277 жыл бұрын
Impressive stuff :D I think for better distance calculation, you should use the height or the area of the objects. The width of a car changes a lot if you rotate it. Maybe the box ratio will help you to detect if you see the side or the front/back.
@novnod7 жыл бұрын
First time commenting on your video although I've been watching for the longest. Keep it up bro.
@RaviPrakash_187 жыл бұрын
All the classes of object that tensorflow can be seen in "models\object_detection\data\mscoco_label_map.pbtxt" file
@poseidon46757 жыл бұрын
AWESOME! I love this concept! I got into tensorflow myself too recently, but this is amazing!
@AnshumanKumar0076 жыл бұрын
I'm building a facial recognition framework using facenet and Mxnet and i thought that was dope until I discovered your channel
@creative.money_eu7 жыл бұрын
Man I love these videos thank you for this, makes me want to buy a new PC!
@PureWaterSolutions7 жыл бұрын
This video is insane! Great work broski! 💻😎✌️
@hakimke25 жыл бұрын
Awesome! I have the same idea on how to do it (in theory) but do not know how to translate it into code. Love your video~
@alokranjan57946 жыл бұрын
@sentdex Could you please share your expertise how we can measure the distance of known object size from the camera. Suppose the dimension of the object is 50*50 mm then how the method you have shown could be utilized to compute the distance. Thank you for your excellent video to the community.
@guardianpresencerobotics77766 жыл бұрын
best thing for the road right now
@suicidalbanananana7 жыл бұрын
Great video series man!
@sushantsawant67035 жыл бұрын
i need one help from you . do you know how to do distance detection with webcam and detected objects
@__Beant__7 жыл бұрын
Absolutely love your videos, Sentdex. I'm currently studying robotics and building an object detection module for a soccer robot, your channel has been invaluable. In a real world environment, the focal length of the camera would also come into play. Would it be possible to let the AI learn with the distance information to improve object avoidance?
@taufeqtauf18177 жыл бұрын
So if our oject of interest is a car we can find the average pixel length of a (average)car at the closest distance and with respect to the area size of the detection box determine its nearness. If we use the same approach for a truck as well we can have a distinct comparison neasure for a car and a truck's distance which also differ in screen area and detection box area size. Although, neverltheless it doesnt seem to be a problem. The same approach can also be used to better judge pedestrian distances to set a safe distance. What do you think of this ?
@rutujashah39106 жыл бұрын
Hii sir, I am running this code on the spyder. so there is error in grabsceen which is already included in my library.so how can I remove this error?
@gugasevero764 жыл бұрын
Hi there! If I have a camera, pointing to diagonal of one street, showing/viewing cars parked, in a straight line, right? So, how can I meansure distances between cars? Thank you so much for your attention.
@karkinissan7 жыл бұрын
For anyone else who can't find the file. The label file is in the "data" folder. It's called "mscoco_label_map.pbtxt". I was shouting at the screen.
@FransHenskens7 жыл бұрын
AI learning algorithms are a chaotic system. They're very sensitive to input data. If you know the best way to filter out noise before applying one, you'll get surprisingly good results. Add an image scaling layer in the net for a few different zoom levels, applying it with an appropriate overlap (1/2 largest relevant object). If you can train these on a frame-by-frame basis for correct object detection (using obj location/width/angle logging) and pass that to a classical neural net (or SVM!) the output should be pretty great :) Or, if you can find think of sensible contrast functions between the output of these, using CNN would be better?
@alokranjan57946 жыл бұрын
Hey, Thanks for the great video and details. Just curious how you selected the size for different objects in your tutorial for distance estimation. You have chosen car=3 , truck=6 and bus= 8. Is this is related to the index in the .pbtxt file? Could you please share your suggestion, Suppose i know the dimension of the object then how can I estimate the distance from the camera or from my place? For example, a logo of a company sign is having 50 mm diameter then how to know the distance from my standing place? How can i translate the index (in your video say 0.2) into meters?
@sankalpsahu21544 жыл бұрын
the code is running but it's not showing the video on the screen the cv2.imshow('window', screen).
@balajichetty3056 жыл бұрын
Hi Sentdex, your tutorials are awesome. Can you describe, and help me to figure out how to control the drawing of bounding boxes only for certain classes? Thanks
@Zhak76 жыл бұрын
You sould probably take into account the current speed of the vehicle your character is driving and also the rear lights of other cars. You could then predict when to slow down and even stop before hitting any car at all
@fatihelibol583 жыл бұрын
Hey, please help me. How can I print the found objects on the console?
@pringlesss77014 жыл бұрын
at 3:05 it's : [y_min, x_min, y_max, x_max]
@projectwork45684 жыл бұрын
Any V2V or V2X video you uploading in coming days?
@abir92334 жыл бұрын
Thank you for the amazing Tutorial ! i just have a question can we replace screen grabbing by an MP4 video?
@tuhinmukherjee81413 жыл бұрын
Yes ofc, any video stream would work as it's just extracting pictures frame by frame.
@milaandahiya7 жыл бұрын
What OS are you running? And if you are running Ubuntu do you use Wine to run GTA? Or do you dual boot with Windows?
@sentdex7 жыл бұрын
100% is happening in Windows.
@illank867 жыл бұрын
can you use tensorflow to detect traffic, if its jammed or not? could you create tutorial for that... i have following the tutotrial till this video, but has no idea how to implement it to check the traffic
@vicveesi7 жыл бұрын
What are the index that he used in "classes[0][i]". Where can I learn a theory about it?
@malotabi5949 Жыл бұрын
thisis amazing can we consier the warning as an attack?
@shivrajkumar89144 жыл бұрын
i did tensorflow realtime object detection and its working fine for custom dataset by looking your videos, now i want to try please tell me how can we get that video and where u will store that video in object detection folder
@Jakeh0776 жыл бұрын
Hi sentdex, do you happen to know the units that the distance output is in, or how to get it into meters? Thanks and i really enjoy watching your videos!
@sentdex6 жыл бұрын
well, it's just approximate. I wrote it. I forget which unit of measure I was using. I just eye-balled it. The scale isn't perfect tho. It's just all based on how wide the car is in pixels. You're probably better off just using the raw number, rather than doing some conversion, but I liked the idea of converting it to something reasonably close to feet or meters.
@Jakeh0776 жыл бұрын
Ah ok gotcha. Thanks for the fast reply! Also just wanted to say thanks for your videos. I’m a biomedical engineering student and for my senior design project we are making a navigation assistant for the visually impaired. And we have used several of your tutorials to help learn about tensorflow for our project!
@codetech60997 жыл бұрын
Hey there i am watching your video series from a month and getting knowledge about python programming.(i have subscribed your channel). My question is that Can you suggest some books for python programming?
@HarshSingh-np3ts6 жыл бұрын
Hey sentdex, Can you help me with increasing the processing speed of the frames or on what factors does the processing (marking objects,warnings etc.) depends on? And thanks your videos are great :)
@kasunhewagama64533 жыл бұрын
In the code, you take car == 3, bus == 6, and truck == 8. from where did you take that measure? How did you do that? any logic?
@vismayamenon17194 жыл бұрын
Is it possible to apply this method on any other object detection problem ? for example person distances and collision warning ?
@luthontantiso77475 жыл бұрын
how do you only detect cars in image or video with detecting other objects?
@lokilocus76207 жыл бұрын
could u make a video series about writing lexers, parsers etc for creating a programming language?
@jayakumarj71736 жыл бұрын
Is it Possible to implement this idea to real life....?
@default6324 жыл бұрын
Yes. But it'll be much more expensive to test and train.
@durgaprasadreddy54487 жыл бұрын
hey sentdex can we do this for autonomous navigation of drones???
@nate00314 жыл бұрын
What kind of hardware is needed to run tensorflow like this?
@mehranali33105 жыл бұрын
is that same work for the object detection and distance from camera to objects
@muhammadazmizainudin46247 жыл бұрын
is it possible to get the speed of vehicle approaching you since you already know the distance of the vehicle?
@ahmadrashed16726 жыл бұрын
many thanks to you keep going
@MSalman17 жыл бұрын
Just amazing work Harrison! Any suggestions for people who cant use GTA due to hardware requirements?
@sentdex7 жыл бұрын
Try san andreas. No idea i fit works, but I suspect it would.
@arjunkrishna88737 жыл бұрын
few of us did a project on nfs last year using similar methods you can try that
@guruvishnu117 жыл бұрын
Hi Sentdex, Thanks a lot for a Superb Idea and Code, Can you please let me know, if there is any chance of adding an mp4 file as input, instead of screen grab?
@StrawHatLuffy7505 жыл бұрын
use cv2.videocapture() .could be done with opencv.
@sumitpatnaik70575 жыл бұрын
Can I use tensorflow lite in raspberry pi
@amitjajoo95104 жыл бұрын
can u explain which distance is this
@prabhu02256 жыл бұрын
First of all thank you soo much for all this video ..I would like to do a project like ..finding the object and finding the object distance from the camera ..could you please tell me ??how I can
@alexboitor56997 жыл бұрын
does somebody know how to block tensorflow from detecting certain objects with this model? i mean, i want to use this model, but i don't need cats, dogs, or potted plants for example (and i would guess removing items would speed up the processing). I watched the tutorial series to create a custom model, but i'm curious if i could remove some items from this one. Thanks in advance!! P.S @sentdex you're the best!
@garceling3 жыл бұрын
Can I used the same code in real footage on the road
@WorldWithAravind5 жыл бұрын
how do if check the distance between two object
@nadarbzh7 жыл бұрын
Im really intrested in self driving car and it would be cool if you could train a model that drives like a real self driving car. Maybe from point a to b via maps if and following the rules as good as possible;)
@sentdex7 жыл бұрын
Charles drives point to point already www.twitch.tv/sentdex ... but he doesn't follow rules. Working currently on a way to get a few versions of charles, one that does try to follow rules, but we'll see.
@nadarbzh7 жыл бұрын
Thanks for the answer. I've seen the navigation via maps but it would be super cool to see it in combination with following rules and how good this model would work compared to the big players like waymo ;) Btw love your tutorials!
@raghuprakashreddy17945 жыл бұрын
Please share the link for dowloading the sample gameplay
@alexanderparker19927 жыл бұрын
Would it be possible to use this to make additional training data for your original self driving implementation and improve it?
@fleecemaster7 жыл бұрын
I'm pretty sure that his plan is to integrate it, which I'm very excited about :)
@CreativeShorts753 жыл бұрын
Is this a self driving car ,I mean does it move automaticallly?
@ahmadmaher53266 жыл бұрын
Bro , How we can crop objects from a video based on scores . Is this possible?
@Salman-xz4km7 жыл бұрын
Hey !!! I just want to know can we use this API to train our model in realistic environment like GTA V instead of programming our own customize objection code of every different game or environment ... So that we can make our Ai more generalized (some algorithm of different games).... Thanks in advance
@vishavsingh37175 жыл бұрын
How much Fps are you getting and can your algorithm works for detecting an obstacle at about 10m away from car
@Jayjlow7 жыл бұрын
How much of an impact (frame rate and detection wise) would it be if you detected every 3rd or 5th frame instead of every frame?
@sentdex7 жыл бұрын
It seems to me to make more sense to just allocate as much vram as you want to the app, and let it maximize that. It's already not doing every frame.
@JohnDoe-kr9zh7 жыл бұрын
your the man!
@thrupthirrai93095 жыл бұрын
i did not get any error nor i dint get any output .. please help me
@clinton119947 жыл бұрын
Hi . What minimum system configuration do you suggest for deep learning ?
@sentdex7 жыл бұрын
8GB of vram NVIDIA GPU, 16GB+ RAM. i7 processor, or you can use a cloud provider for the DL training. I've been using paperspace on the side when my GPUs are occupied. I covered them here: kzbin.info/www/bejne/boOtiWuNg8iLr6s
@pottingmix37337 жыл бұрын
I use a gtx 750 ti to run the models and was using google cloud for training but paper space is better :P
@clinton119947 жыл бұрын
Thank you :)
@RahulSahu-hp1kz Жыл бұрын
How can i do it on an video or webcam
@noway28317 жыл бұрын
I am a teenager and really want to learn how to code properly, in python and C# instead of manipulating html pages; how should I go about learning the languages?
@paparibaruah97435 жыл бұрын
Can plz tell me how we import n setup the utils
@TAWPhotography7 жыл бұрын
Amazing job. God Bless.
@ronaldmiranda6926 жыл бұрын
Nice work. I have a question , Can I implement this in real time ?
@HarshSingh-np3ts6 жыл бұрын
I want to know that too. Please let me know if you got the answer :)
@DeepakKumar-uz4xy4 жыл бұрын
How to get real time distance measurements from direct web cam
@santoshmishra95735 жыл бұрын
I got lots of error in module even after instalation can u suggest link how to correct it
@muhammadsaad234 жыл бұрын
You give me a FYP Idea bro thanks
@sisyphus68007 жыл бұрын
Do you think you can do a qpython3 on Android tutorial??
@marr737 жыл бұрын
hello, what specs of laptop / pc should i buy for machine learning?
@joemAwesomeMan7 жыл бұрын
Which ever you want.
@gustavovelasquez10663 жыл бұрын
19:10 LMFAOOOO since this guy is no longer with us
@arielablan59607 жыл бұрын
what algorithms have you used for this?
@sentdex7 жыл бұрын
Everything has been covered in tutorials up to this. Many algorithms have been used.
@bendekoumamina74385 жыл бұрын
do you have the code source of the vehicle detection in android?
@PremKumar-rp6ko4 жыл бұрын
can you share the code for this one for my study purpose
@thorodinson74677 жыл бұрын
Can you make a video on installing modules? I can't do many of these things because my python doesn't like installing them