02 / Computer vision · Deep learning
RealWaste Image Classifier.
Compared transfer-learning models for classifying waste images across nine categories.
01 / THE PROBLEM
Context matters.
Waste classification is a practical image-recognition problem, but overall accuracy can hide uneven performance across categories.
02 / APPROACH
From idea to workflow.
Worked with 4,752 images across nine classes and compared ResNet50 with InceptionV3. The reported accuracy was 88.9% for ResNet50 and 89.19% for InceptionV3. The final write-up should pair these figures with the test setup and per-class results.
03 / TOOLS & METHODS
The stack.
04 / EVIDENCE
Show the work.
Add a real architecture diagram, dashboard screenshot, or model evaluation output here. Explain what the evidence shows and how it was produced.
Add a real diagram, chart, screenshot, or model output here.
05 / REFLECTION
What I learned.
Describe one technical decision, one challenge, and one limitation or next step. Specific details make the work easier to assess.
Confirm the final metric, test split, and repository URL before publishing. A confusion matrix and per-class metrics would make the comparison easier to evaluate.