ijyliu / computer-vision-project

Using classical and neural image embeddings and finetuned end-to-end networks to achieve top-tier performance on a vehicle type classification task. Containerized and deployed model as a web app
https://cv-web-app-3m4f2rmfzq-uc.a.run.app/
0 stars 0 forks source link

HSV Plots for Report #60

Open ijyliu opened 2 months ago

ijyliu commented 2 months ago

image

ijyliu commented 2 months ago

@ashutoshtiwari13 please still try to do the average plots, we might include them

Combine all_features/train and all_features/test into one dataframe, then collapse all of the HSV columns to their averages by class. Separate out the hue, saturation, and value components according to how they were constructed in the HSV embedding creation function. Use your average hue vector, saturation vector, value vector for each class to create 4 plots/histograms like those in the above image, one for each class, that are the averages.

ashutoshtiwari13 commented 2 months ago

done and psuhed the code in the Code/Feature Construction/Color Histograms/avg-HSV-plot.ipynb

ashutoshtiwari13 commented 2 months ago

Done!

ijyliu commented 2 months ago

can you try combining each row, so we end up with 4 plots? you will have to set some alpha

ijyliu commented 2 months ago

task from google doc: set a fixed y-axis across all the plots so we can compare values