Hi all! Ive made a github pages to show case my work. Created a weasel or pine marten classifier also recreated the pet classifier. Trying to make a Flower classifier at the moment using the Oxford Flower data set.
link to pages: https://shab00.github.io/
I am a new guy about deep learning, i just finished the Lesson 1, but when i run the code from duckduckgo_search import ddg_images, there is an ImportError: cannot import name ‘ddg_images’ from ‘duckduckgo_search’ (/opt/conda/lib/python3.10/site-packages/duckduckgo_search/init.py) .
I don’t know how to do it. Could anyone can help me?
I have fixed the problem. The ddg_images is not exist. I have changed it to search_images_ddg, and the whole world is quiet.
def search_images(term, max_images=30):
print(f"Searching for '{term}'")
return L(search_images_ddg(term, max_images=max_images))
# return L(ddg_images(term, max_results=max_images)).itemgot('image')
Thank you for your tip, the ddg_images doesn’t exist, should use search_images_ddg method. Thank you.
I had a similar issue and finally fixed it by modifying the import (duckduckgo_search was not working for me…). Here is the snap of the code in case it’s useful to someone:
from fastbook import search_images_ddg
from fastcore.all import *
def search_images(term, max_images=30):
print(f"Searching for '{term}'")
return L(search_images_ddg(term, max_images=max_images))
Hi all, here is a screenshot of my first model. I decided to go for a “string of pearls” vs “string of turtle” plants classifier. I selected the photos myself, ensuring I had a huge variety for both categories, and I adapted the code. It was quite fun 
.
Hey everyone! I recently finished Lesson 1 and made this fungi image classifier for practice. I was able to get an accuracy of ~93% compared to the original paper’s ~85%. I’d love to hear your thoughts.
Really loving this course (and the fastai library). Looking forward to the rest of it.
In fast.ai, I completed one session in which I learned about neural networks and categorical classifications using the Birds or Not example. Using fast.ai, I developed a small project that predicts whether a vehicle is damaged. To predict the results, I used a great deal of images of Indian vehicles.
Kaggle Link: Project Link
Thank you, i used the same method search_images_ddg and it worked.
Your model is overfitting. The error rate increases for the fourth time and remains stable afterwards while the training loss continues to decrease, performing well. Therefore, it’s generally fine to fine-tune the model three times, or to incorporate methods for dealing with overfitting during this process.
I have tried your method, but still got an error and couldn’t import.
Another method from fastbook import search_images_ddg works well for me.
You gotta do !pip install duckduckgo-search 
before importing the module
Buy anyway Happy you got it figured out…
Hi all, Really enjoying the course so far. I decided to create an Adidas, Nike and Puma sneaker classifier.
Hey everyone. I’ve been following the CUDA MODE lecture series learning about GPU programming. It’s a bit of an advanced topic, but as usual, Jeremy has some fantastic and very accessible notebooks (lecture 3 and 5) to help you get started.
I recently ran into a problem that was well suited for GPU acceleration so I tried my hand at writing my own custom CUDA kernels to solve it. I was able to achieve a 8.4 Million times improvement over a naive pandas implementation and more interestingly to me a 45 times speedup over broadcasting tensors on GPU with PyTorch, which was quite surprising to me. Here is the link to my blog as well as the link to the tweet.
Hi Carlos, Great work!
Could you please check hugging face link? It shows me error running the model European Tree Classifier - a Hugging Face Space by CarlosSMWolff
I have been wondering about this question for a long time, and finally, thanks to the (Lesson 1) course from fastai, I was able to get an answer to it.
Hi everyone, I’ve just started learning DL from Fast.ai and would to share a simple model to classify a Plant and a Tree. Here’s the Notebook:
Please feel free to share your thoughts in the comments of the Notebook. Just a beginner here who’ll be actively learning DL and would love to connect with other learners.
Hi everyone, I’ve just finished two classes of this course. I built a model to decode the baseball secret code of stealing. I know nothing about baseball but I watched Mark Rober’s youtube video:
I randomly generated a int number between 1 and 4 for 5 times. That int number will be 1, while the other int numbers from 1 to 4 will be 0. I transformed this into a colorbar plot. I catogarize these photos into steal and not steal based on one little secret. I tried to use photo classifier to identify which is which. I fine-tuned it for 10 times and the result is 100% accuracy!!
I don’t know how baseball works either. But I’m Interested in what the secret is though 
