Student Assistant for Benchmarking NeRF and 3D Semantics Methods

11.03.2024, Studentische Hilfskräfte, Praktikantenstellen, Studienarbeiten

Want to put your deep learning skills to use with the latest NeRF papers like ReconFusion and ZipNeRF? Or more interested in 3D understanding papers like Mask3D and SPFormer?

The 3D Understanding Lab at TUM is looking for a Hiwi to set up and add these methods to the ScanNet++ Benchmarks.

This is a great opportunity to level up your deep learning skills and get ins ights into cutting edge computer vision research.

• Literature survey of most relevant novel view synthesis methods - estimate required compute, training time and expected results
• Setup code from the original paper and adapt it to the ScanNet++ dataset
• Setup logging of metrics and results
• Monitor and report results, add them to our benchmark website

• Experience in working with complex computer vision/deep learning codebases
• Strong skills in debugging Python / Pytorch, tuning hyperparameters, visualizing results
• Structured method of working and good communication skills
• Relevant coursework and project experience such as I2DL, ADL4CV, 3D Scanning and Motion Capture

Kontakt: yuehcheng.liu@tum.de, chandan.yeshwanth@tum.de