Computer Vision Engineer (US)

How do you decide between a two-stage object detector and a single-stage detector for a given application?

I use a single-stage detector when inference speed matters more, like a real-time application, since two-stage detectors are generally more accurate but slower. If the application can tolerate more latency and accuracy is the priority, like offline batch processing, a two-stage approach is often worth the extra cost.

What's your approach to handling class imbalance in a training dataset for image classification?

I use a combination of class-weighted loss functions and targeted data augmentation for underrepresented classes, rather than relying on oversampling alone, since naive oversampling can lead to overfitting on the minority class. I also check whether the imbalance reflects real-world distribution, since forcing artificial balance can hurt real-world performance.

How do you evaluate whether a model is overfitting to specific characteristics of your training data, like lighting or camera angle?

I test on a held-out set that deliberately varies those conditions rather than just a random split of the same distribution, since a random split can share the same biases as training data. If performance drops significantly under different conditions, that's a signal the model learned spurious correlations rather than the actual task.

What preprocessing steps do you typically apply before feeding images into a vision model, and how do you decide what's necessary?

I match preprocessing to what the model architecture expects and what the deployment environment will actually produce, rather than applying a standard pipeline by default. Normalization and resizing are almost always necessary, but more aggressive preprocessing, like heavy denoising, depends on the actual noise characteristics of the input data.