Sign language image classification – Fine-tuning MobileNet with Keras

Spread the love

Sign language image classification  - Fine-tuning MobileNet with Keras

In this video, we’ll be building on what we’ve learned about MobileNet and the techniques we’ve used for fine-tuning to fine-tune MobileNet on a custom image data set that does not have classes similar to the ImageNet classes it was originally trained on.

We saw in the last video how well our fine-tuned MobileNet model performed on classifying images of cats and dogs. We noted, however, that many types of cat and dog breeds were included in the ImageNet data set that MobileNet was originally trained on, so we didn’t have to do too much tuning for our model to perform well on classifying the images.

Now, we’re going to see what we need to do to fine-tune MobileNet on a completely new type of data set, which doesn’t contain classes similar to those included in ImageNet. This data set is images of sign language digits. There are ten classes, labeled as zero through nine, and each class is made up of images of hands showing the sign for that particular digit.

Code files are available as a perk for the deeplizard hivemind. Check out the details regarding deeplizard perks and rewards at:

Sign Language Digits data set:

Follow deeplizard:

Support collective intelligence, and join the deeplizard hivemind:

Checkout products deeplizard suggests on Amazon:

Support deeplizard by browsing with Brave:

Support deeplizard with crypto:
Bitcoin: 1AFgm3fLTiG5pNPgnfkKdsktgxLCMYpxCN
Litecoin: LTZ2AUGpDmFm85y89PFFvVR5QmfX6Rfzg3
Ether: 0x9105cd0ecbc921ad19f6d5f9dd249735da8269ef

Recommended books on AI:
The Most Human Human: What Artificial Intelligence Teaches Us About Being Alive:
Life 3.0: Being Human in the Age of Artificial Intelligence

Data Science –
Machine Learning –
Keras –

Laser Groove by Kevin MacLeod
Chillin Hard by Kevin MacLeod
Investigations by Kevin MacLeod
Thinking Music by Kevin MacLeod
Licensed under Creative Commons: By Attribution 3.0 License

Author: administrator