This is exciting. I am having trouble with improving the performance of my segmentation models. Looking forwards to Part 2. Thanks for the great work that you are doing.
It is a pretty good and clear video for explaining how to the training network and display intermediate parameters and can learn skills for tuning it in Unet pattern.
Thanks Sir. I cann't wait til you have uploaded the code to the github. I am typing codes from your screen. After a few typo fixes, i can see what you explained, on my screen.
The code is already on Github. Please do not waste time typing code from watching the video screen, not a good use of your time. Just copy the code and spend time in customizing or enhancing it.
Since the original images might be divided into 256x256 patches, when training the Encoder do you recommend also including the patches that doesn't include the region-of-interest. What about when training the Decoder. Also, whats the effect of "Smooth blending" of patches on the Encoder vs Decoder training.
thank you for this amazing content, do u have any idea about bushfire satellite datasets? ..... I can only find satellite imagery of the fires in csv. in csv files i can't see semantic segmentation as a result !!!! ,and are fires possible for semantic segmentation ?
line 77 : model_for_visualization = Model(inputs = my_model.input, outputs = outputs) i am getting error as name 'Model' is not defined please help me in this
I have done a few videos on this topic where we can use pretrained CNN as feature extractor and traditional approach for classification (e.g., Random Forest or SVM or XGBoost, etc.). Here is a link to one of those videos: ua-cam.com/video/2miw-69Xb0g/v-deo.html
You can define U-net for any shape, not just for square aspect ratios. Similarly you can define any patch size for patchify. Find a patch size that works for both dimensions.
@@DigitalSreeni one more doubt while training the the unet model for large images do we have to divide both the original image and the mask into patches and then give these patches as inputs to unet with their corresponding mask patches ?
@DigitalSreeni one more question if you're free answer me without fail.... Case 1: I trained a yolo v4 model with two classes. Now i has to train same model with adding another two classes. Train the model without losses of previous two classes weight...is this possible. My answer : reserving extra node in output layer. Can i do this? Your answer for case1 : Case 2 : Dataset description: 4k images with two classes and balanced classes. Using this data set i trained two model using tiny-yolov4. My question is: Model 1 : trained all 4k images. 20k max_batches . getting 84% accuracy avg loss 0.12xxx Model 2: Cycle 1 :i trained 3k images with 20k max_batch getting 94% accuracy. Cycle 2 : i trained 1k images with 20k max batch using last weight of cycle 1. After completion i am getting 94% accuracy. And avg loss 0.0xx. My question is both the model i trained same set of images why result is different. Training small set of image is good? Even though i increased model 20k+20k max batch thare is no improvement. Note: cfg file are same for both model. Thanks
Thanks so much sir for your great tutorials. I'd really appreciate it. I want to autoencode an image into a float number (not classification). For example, taking an microscopic image and labeling it with elastic properties. Can you help me in this regard
I wish I had time to help with personal projects but unfortunately my day job takes up all the time. I can offer my point of view if you have any questions.
This is exciting. I am having trouble with improving the performance of my segmentation models. Looking forwards to Part 2. Thanks for the great work that you are doing.
It is a pretty good and clear video for explaining how to the training network and display intermediate parameters and can learn skills for tuning it in Unet pattern.
Thank you
I wanted to take a moment to reach out and thank you for creating such informative and helpful videos.
You are the on of the best classic content creator for computer vision. Kindly do it for nlp as well.
Just watched this and without any hesitation clicked the subscribed and like button. YOU are absolutely great. Keep it up.
Welcome aboard!
Thanks Sir. I cann't wait til you have uploaded the code to the github. I am typing codes from your screen. After a few typo fixes, i can see what you explained, on my screen.
The code is already on Github. Please do not waste time typing code from watching the video screen, not a good use of your time. Just copy the code and spend time in customizing or enhancing it.
@@DigitalSreeni ys I overlooked but I am still in beginner learner stage in Deep learning and I grab more when I do type-and-learn. Thanks again Sir.
An extremely underrated channel
Thank you very much :)
Great concept and this blew my mind. Never knew you could use pre-trained weights other than the ones trained on datasets like imagenet.
please start some series on mask rcnn. Thanks for your contribution to this computer vision world.
Since the original images might be divided into 256x256 patches, when training the Encoder do you recommend also including the patches that doesn't include the region-of-interest. What about when training the Decoder. Also, whats the effect of "Smooth blending" of patches on the Encoder vs Decoder training.
Is there possible to add blocks in UNet's encoder and decoder to add new image restoration algorithm like ( grey world, retinex )
thank you for this amazing content, do u have any idea about bushfire
satellite datasets? ..... I can only find satellite imagery of the fires
in csv.
in csv files i can't see semantic segmentation as a result !!!! ,and are fires possible for semantic segmentation ?
line 77 : model_for_visualization = Model(inputs = my_model.input, outputs = outputs)
i am getting error as
name 'Model' is not defined
please help me in this
How much RAM your device have ? Mine crashed while trying to running this autoencoder.
When I recorded the video, I believe I had 4GB GPU. Now I have 16GB. If you are serious about deep learning you need at least 16GB GPU, ideally 32GB.
hi sir , i want to ask ,could we use CNN with HOG and linear SVM classifier for object detection?
I have done a few videos on this topic where we can use pretrained CNN as feature extractor and traditional approach for classification (e.g., Random Forest or SVM or XGBoost, etc.). Here is a link to one of those videos: ua-cam.com/video/2miw-69Xb0g/v-deo.html
How can we apply u net segmentation on images which have an odd shape such as 450x1450 how do we use pachify in this case ?
You can define U-net for any shape, not just for square aspect ratios. Similarly you can define any patch size for patchify. Find a patch size that works for both dimensions.
@@DigitalSreeni one more doubt while training the the unet model for large images do we have to divide both the original image and the mask into patches and then give these patches as inputs to unet with their corresponding mask patches ?
Amazingly good, wonderful. Thanks.
Thank you too!
Would you please make atutorial about mask RCNN
Sir, can we use this concept for image classification problem?
Yes, of course. For classification you don't even need the decoder, just take the trained encoder and add a classifier to it.
@DigitalSreeni one more question if you're free answer me without fail....
Case 1: I trained a yolo v4 model with two classes. Now i has to train same model with adding another two classes. Train the model without losses of previous two classes weight...is this possible.
My answer : reserving extra node in output layer. Can i do this?
Your answer for case1 :
Case 2 :
Dataset description: 4k images with two classes and balanced classes. Using this data set i trained two model using tiny-yolov4.
My question is:
Model 1 : trained all 4k images. 20k max_batches . getting 84% accuracy avg loss 0.12xxx
Model 2:
Cycle 1 :i trained 3k images with 20k max_batch getting 94% accuracy.
Cycle 2 : i trained 1k images with 20k max batch using last weight of cycle 1. After completion i am getting 94% accuracy. And avg loss 0.0xx.
My question is both the model i trained same set of images why result is different. Training small set of image is good?
Even though i increased model 20k+20k max batch thare is no improvement.
Note: cfg file are same for both model.
Thanks
thanks dear sir, could you please upload next video.
Nice video sreeni, send me link for the dataset.
This video uses a single image of Monalisa. You can easily find similar (or the same) image by searching online.
Thanks so much sir for your great tutorials. I'd really appreciate it.
I want to autoencode an image into a float number (not classification). For example, taking an microscopic image and labeling it with elastic properties. Can you help me in this regard
I wish I had time to help with personal projects but unfortunately my day job takes up all the time. I can offer my point of view if you have any questions.