EfficientNet on Custom Dataset | Image Classification Using EfficientNet

EfficientNet Practical Implementation On Custom Dataset

EfficientNet is a convolutional neural network architecture and scaling method that uniformly scales all dimensions of depth/width/resolution using a compound coefficient.

The researchers used the compound scaling method to scale the dimensions of the network. The applied grid search strategy to find the relationship between the different scaling dimensions of the baseline network under a fixed resource constraint. Using this strategy, the could find the appropriate scaling coefficients for each of the dimensions to be scaled-up. Using these coefficients, the baseline network was scaled by the desired size.

What does scaling mean in the context of CNNs?
There are three scaling dimensions of a CNN: depth, width, and resolution.
Depth simply means how deep the networks is which is equivalent to the number of layers in it.

Width simply means how wide the network is. One measure of width, for example, is the number of channels in a Conv layer
Resolution is simply the image resolution that is being passed to a CNN.
Compound scaling:
Compound scaling method uses a compound co-efficient ø to scale width, depth, and resolution together.

#ai #computervision #artificialintelligence #artificialintelligence #deeplearning #python

[tryhardcsnoob]

Thank you so much! I have been struggling really hard with university assignments and this video really helped me a lot!

[mayuratapkire]

Thanks. In depth explanation. Cleared all my doubts.

[hippasus]

Thanks soo much, was struggling to understand using of such models since so long, really do appreciate your work, again thanks a lot

[sudheer]

How to plot Confusion matrix in the same code

[mikhaelteofilus]

Thank you Mam, great video, but
anyone see the link of next video about 'Transfer Learing with EfficientB0' ?

[ccsushant]

Awesome explanation. Simple and Clearly explained. Thankyou so much Mam..Looking forward for more videos..

[_ifly]

amazing as alway keep this way of teaching ❤

[madhurimasarkar]

Thanks for uploading this. I am looking for this.

[shekharkumar]

Very informative video series on efficient net.Arohi Mam, can we use efficient net on realtime videos to detect a single object and how?

[seraphritt]

Thank you so much for the explanation! +1 Subscriber keep up the good work!

[Kishi]

Very educative and mind blowing lesson..But if someone do not have GPU, How can he test that of GPU ma ?

[thestoryteller]

Thank you for such a productive content. I got an error while running EfficientNetB4,
Colab session ends as the RAM exceeds. I don't know how to resolve this issue.

[vinaykumar-cqln]

good explanation but i have one doubt when input images are in different sizes can i make the directly required image size as 224x224 for efficientnetB0 and the remaining model requires a different image size. it may lead to loss of information from original images

[ayushkurlekar]

Nice video, kindly add video on fine tuning in this case where the data is small.

[shantilatanayak]

Why you didn't upload the transfer learning video for EfficientNet as you mentioned in the video??? Please upload!!!

[iftikharahmad]

Great explanation Aarohi Madam, Can you please explain YoloR With a custom dataset, please?

[robr]

Thanks for the video. Getting this error: TypeError: Failed to convert elements of SparseTensor

got this error at last stage
hist = model.fit(train_x, train_y, epochs=30, verbose=2)
at this line

Any thoughts?

[shivamsingh-fnvz]

well great video, but i have a doubt i have read in many research papers that one how encoding is not a great option to use when working with mullti class classification . So if i use tensorflows to_categorical feature how is it different from one Hot encoder

[CodingMaverick]

Mam thanks for your explanation . As you said, i augmented the data and increased my test accuracy to 94%, but the problem is, by giving any unseen data, it is not predicting good, may i have some suggestions or solutions regarding my issues?

[rama_bhuyan]

Hello
in general how many images should be taken per class for a decent model ?
Thank you for such videos, hoping for more such contents