I am loving this fast ai course by Jermy I have made my own Drone classifier using fast ai and have deployed it on render.Drone Classifier Please take a look int it 
My Drone Classifier take in drone image and classifies whether it is
1)Singlerotor Drone
2)Multirotor Drone
3)Fixedwing Drone
And Outputs the class of the Drone
1 Like
Hi Ungast hope your having a beautiful day!
I had a look at your app and really like the way you have designed an app with a little privacy in mind. It makes a refreshing change from the majority of apps that are in reallity nothing but data gathering apps!
Cheers mrfabulous1 
2 Likes
Taking inspiration from the part-1 course I have tried to explain the high level structure of a CNN in a beginner friendly way. I have tried not to delve into all the minute technical details of a CNN. The blog can be found here
1 Like
Hi sapal6 Hope all is well!
A wonderful clear and concise explanation.
I wish all writing was as clear and concise.
Cheers mrfabulous1
1 Like
Thanks a lot. I hope that I was able to convey the message in a proper way.
1 Like
I have created a model using fast ai for plant disease detection. With just fine tuning the learning rate i achieved an accuracy of 99.1 percent while the state of art is 97 %. I did not even use the deeper networks like resnet 50. i just used resent 34.
2 Likes
Are you using the Plant village dataset? Every paper I saw has a 99+% accuracy rate.
Nope it’s the dataset from kaggle. It’s highest accuracy is reported as 97.4 .
Btw do you know how can I get access to plant village dataset I came to know it’s no longer public . I mailed them regarding the dataset but no reply yet .it would be helpful if you can point me to the dataset
You can get it from Kaggle https://www.kaggle.com/emmarex/plantdisease
Is it the complete dataset ??!! I can find only pepper potato and tomato.
I deleted this post. But I decided that I should still need to post it because it is a pretty interesting project.
It is a recoloring, but the model focus more on recoloring the skin in the style and color we wants. Dataset has only 200 images, so It sometimes changes the color of hair, except it is something like black hair. In fact, the model was trained that can only focus on skin.
Please note that this image are from the validation set, so it should count as an unseen image.
I will post more images processed by this model.
After
Before
1 Like
Titanic: Machine Learning from Disaster.
It is a model that using ship passenger’s information to determine how much chance that passenger will survive or not.
It was my first time training with tabular data with 900 examples of data for training and only got 72-75% of accuracy.
And it is my first time trying to beat people on kaggle and ended up like ranked in 15030 on Kaggle
I think for a stupid person like me just started in Deep Learning about 1 to 2 months spend 3 hours doing this is pretty good.
2 Likes
Hi JonathanSum You’re cleverer than you think! 
Cheers mrfabulous1 
1 Like
I think the model is pretty successful, and you will understand it if you watch a lot of anime too.
I think if I build a website for an anime fan, I have no problem determining what anime they like based on their review. In addition, I can also group out the taste of anime.
One more thing, I deleted the Anime reviews that were not reviewed. Maybe your favorite one was in there. But what should I do? Should I put them 2-star review because it is not as bad as one star? if you can give me a good suggestion, I will train it again like 4 hours long.
For the whole map, I can tell the upper left region are mostly about alone, violence, sexy. But the middle left is mostly about the team, Clan(Group), decent young Anime. The upper right region is about the monster tale or strange thing. For example Bakemonogatari and the headless biker. The middle right is the region that I don’t like and I don’t know why. Thus, I don’t think I can not really explain it. For the lowerest part is much more an anime for female.
a) plantvillage_deeplearning_paper_dataset/color · master · h / leaf-disease-plant-village · GitLab
b) PlantVillage-Dataset/raw/color at master · spMohanty/PlantVillage-Dataset · GitHub
This also has all plants ( slightly modified )
c) If you have access to crowai platform ( registrations are closed now ) , then you can download it from there
https://www.crowdai.org/challenges/plantvillage-disease-classification-challenge/dataset_files
e) This link
https://zenodo.org/record/1204914/files/plantvillage_deeplearning_paper_dataset.7z?download=1
e) Alternatively , you can write to the maintainer of the dataset and he will provide you with a link to the dropbox
Hi everyone! So I was trying to make some interesting classifier and I decided to make a classifier that classifies an image of guitar into one of Acoustic, Electric, Archtop, resonator and double-neck. Here is the google colab notebook:
https://colab.research.google.com/drive/1MLBWSYBRqhLzLj-IooyxthqnRIXoJ14M
Please suggest any ideas that I can improve my work with
Hey everyone. 
Let me share a classifier with you that distinguishes between seven different herbs. I wrote a tutorial that sums up the first three lessons of part 1 and I’m sharing the friends link to it (so you don’t need to pay anything to have a look…)
In the article you’ll also find a link to a starter kernel, for those who are interested. Let me know what you think. By the way: I’m so glad that fastai exists! 
1 Like
I build sorters.
I have two of them by now - LEGO sorter and Magic the Gathering cards sorter.
The LEGO sorter sorts the parts by mold or color. It finally uses a set of standard single label CNNs based on Resnet34. The CNNs are organized hierarchically - first sorts to basic categories (brick, plate, technic, slope etc.), other networks do the more specific sorting of higher category. I have a data of cca 150 different molds by now. I have at least 500 images for most of the molds.
I tried to build just one multilabel CNN to cover all the hierarchy, but with no big success. The common categories were successful, the detailed didn’t learn well (obviously - much smaller dataset).
The MtG sorter originally used the simmilar hierarchical approach - and it worked quite well. I tried a single CNN, too, but with no success - 50 000 categories were too much for resnet34. Even if I were successfull, I’d have to rebuild the CNN with every new card.
Finally I adapted the @radek 's siamese network from Whale competition. It still has problems to distinguish between the almost the same cards sometimes, but works quite well. The controlling software is connected to the current card prices, so I can sort out the cards worth of selling.
The dataset contains one image per card.
You can see both machines on the video from LEGO exhibition in Olomouc, Czech.
There are a few more videos in my channel:
https://www.youtube.com/channel/UCfc7oHyDpceKFabTuM9Hzew/
11 Likes
Hello!
I finished lesson 2, and have to say this course is the best course I have ever encountered, hands down. The top-down approach is amazing. Being able to train and deploy deep networks from day 1 is amazing.
I decided to make a model that my friends and family would be able to have some fun with, so I googled pictures of attractive and unattractive people, and made a cnn classifier to try to be able to find the difference between the two, and after cleaning up the data (a lot, the searches provided a very noisy dataset), my model achieved about 94% accuracy. I deployed it as a web app on Render using the bear classification template you provided, and my friends are having lots of fun with it!
Here is the link: hot-or-not.onrender.com
Thank you so much for making this course, and the fastai library!
Here is the link to the notebook, I have no idea how to link it properly, it would be cool if someone could tell me 
https://gist.github.com/Robertleoj/8e64cda6188a7beb993c8e330a28f186