C4W3L01 Object Localization
Vložit
- čas přidán 29. 08. 2024
- Take the Deep Learning Specialization: bit.ly/2IpmuHg
Check out all our courses: www.deeplearni...
Subscribe to The Batch, our weekly newsletter: www.deeplearni...
Follow us:
Twitter: / deeplearningai_
Facebook: / deeplearninghq
Linkedin: / deeplearningai
As soon as I find a video by Prof Andrew on a topic I am looking for, I know this topic is so done for good.
Thanks, Prof for these wonderful lectures. I can't be enough grateful.
pro trick : you can watch series on Flixzone. Been using it for watching lots of of movies lately.
@Oscar Carson yup, been watching on Flixzone} for since december myself :)
@Oscar Carson Definitely, I've been using flixzone} for since november myself :D
@Oscar Carson yea, I have been using flixzone} for months myself =)
@Oscar Carson Definitely, I've been watching on Flixzone} for years myself :D
I am a total beginner(even for python).
I couldn't understand the courses here one month ago. Then I took about 1 month to go around and try most of the popular algorithms examples (with GPU linux server). Then I come back and watch the courses. Now I could be more confident to continue the course journey with Andrew.
Hello brother could you kindly share the resources you learned before taking this course
for object detection usually use 'blob ' , contors to separate objects from background and classify that slits
Since the loss function for the case when y1 = 0 includes only one term in contrast to the case when y1 = 0, isn't it kind of encouraging the network to predict that there is no object(background) over the other cases?
Not really. When calculating the loss, it only ignores bx, by, bh, bw, c if the ground-truth value (y) = 0, not when your prediction (y hat) = 0. So if your model tries to predict y hat = 0 more, the loss function will still consider and compare your predicted bx, by, bh, bw, c with their ground-truth values even though y hat = 0.
really awesome lecture!
Clear and Simple!!! Awesone lecture
Thnaks a lot for useful and easy presentation
In last part of this video, he said we can use softmax, squared error, logistic regression loss. if I use that, I think there will be three different type loss. And then how should I do back prop? Just calculate loss matched each output neuron's loss fucntion?
Have you find the answer?
@@abubakarali6399 I found that yolo outputs multiple tensors. In other word, in last part in yolo there is 3 different type layers (may be more) which is p, box positions, clasees
Waah chacha kya tutorial diye hain!!
bc
Great introduction.
Hi everyone I am confused about the bbox part. How does the feature vector stack in final FC layer spit out some arbitrary 4 number that are bbox parameters even before backpropagation and L2 loss part. Are these initial bbox co-ordinates the one of the feature vector that has the object in it? Lets say in the case of localizing a car the final FC layer before softmax will have learnt high level features like car wheels, windshield etc and at the end these are stacked. Having said the bbox of the whole car will be different than the bbox of high level features like wheel, windshield etc. I am confused in this part of predicting the initial bbox of the whole car even though it might not be accurate initially bathos does the bbox of high level feature vector match the bbox of whole object. correct me if i were wrong somewhere.
Nice and clear explanation
This is very good information & helpful. Thanks.
do u work at google?
@@Nishchay-fk5lr NO, I work with Nagarro
this is super good content thanks so much
What will be bx,by,bh,bw value in the output vector if multiple classes are present in the picture.
This video is more about the basic idea of how people encode the training data so he only talks about the case of 1 class in each image. You will have to duplicate this array to support the multiple classes. He talks more about that in the following videos of the course.
see in next video the next problem is that onlyy!!!!!
great lecture
nice explanation .need to watch agaian
You are consistent. Nice.
bounding box data as input, while training a model is give after convolution operation , am I right ?, I have little confusion . :)
Hi! I am wondering why the background is not included in the vectors!
I looked at it as just training with 3 classes and if it cant detect any of them, then it's a background
awesome sir
Awesome!
wonderful
If there 'n' of objects in an image, then how the softmax output will be? will be same [pc, bx, by, bw,bh,c1,c2,c3]? How the output will be?
c1,c2.. extended to the no. of classes
If there is a pedestrian and car in a frame ??? Is it applicable
makes sense
how we have given input image to each neuron
you feed in pixels to the input neurons
Hi Priyanka, this is a good lecture, but I suggest you start with some that are more fundamental. Here is a nice video: czcams.com/video/2-Ol7ZB0MmU/video.html I really like the way Luis explains things so you may start with a few of them first.
Can you name any training set which has the same classes and bounding boxes values to try this approach?
literally all sensors lidar, radar all of them use this approach!!!
but how do we get bx and by value ?
pixle of middle object / max pixel, both for x and y
how could one program "don't care" as an output of an image which contains no object?
Pay attention at 8:59 . When pc_train==0, the loss function is calculated differently: only the object prediction pc_pred is used. So it really "doesn't care" about the values of bw_pred, bh_pred, etc... as there are not in the formula!
in the formula, "don't care" is defined as the "1| _oobj"
could Pc or C2 be between 0 and 1
Pc will be the probability of having an object or not, so this neuron will work as logistics regression only, hence the output can be betwwen 0 and 1. For classes, the concept is almost same and it's like the output of softmax.
How pc variable know without calculating class labels.Because Andrew say if pc 0 the other variable dont care.But how pc know i am 0 or 1?????????
You can check out my repository over object localization for SINGLE object. It is a ready-to-run repository.
github.com/MuhammedBuyukkinaci/Object-Classification-and-Localization-with-TensorFlow
Gonna make TCAS for blind people
Everyone : Awesome ,very nice , great !!!
Me : That's why he is Andrew