Multi-View Priors for Learning Detectors from Sparse Viewpoint Data

Authors: Bojan Pepik; Michael Stark; Peter Gehler; Bernt Schiele

ICLR 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental As we show in our experiments, this transfer is not only beneficial for detectors based on basic-level category representations, but also enables the robust learning of detectors that represent classes at finer levels of granularity, where training data is typically even scarcer and more unbalanced. As a result, we report largely improved performance in simultaneous 2D object localization and viewpoint estimation on a recent dataset of challenging street scenes.
Researcher Affiliation Academia Bojan Pepik1 Michael Stark1,2 Peter Gehler3 Bernt Schiele1 1Max Planck Institute for Informatics, 2Stanford University, 3Max Planck Institute for Intelligent Systems
Pseudocode No The paper describes mathematical formulations and algorithmic steps in prose but does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks, nor does it present structured steps in a code-like format.
Open Source Code No The paper states that 'The annotations will be made publicly available upon publication' which refers to data, but it does not provide an explicit statement or link for the open-sourcing of the code for the methodology described in the paper.
Open Datasets Yes We perform the analysis on two tasks, 2D bounding box localization and viewpoint estimation on the 3D Object Classes dataset [39] (a widely accepted multiview-benchmark with balanced training and test data from 8 viewpoint bins, for 9 object classes)... Specifically, we focus on the car object class on the KITTI street scene dataset [17] (and the tracking benchmark subset)...
Dataset Splits No The paper describes training and test sets and mentions using 'k training examples per view' and 'bootstrapping' for training source models, but it does not explicitly specify a distinct validation dataset split with percentages or counts for hyperparameter tuning.
Hardware Specification No The paper mentions 'For computational reasons, we restrict ourselves to the root-template-only version of the DPM', but it does not provide any specific details about the hardware (e.g., GPU models, CPU types, memory) used for running the experiments.
Software Dependencies No The paper mentions using 'DPM' and an 'SVM solver' but does not provide specific version numbers for these or any other software dependencies needed to replicate the experiments.
Experiment Setup Yes In all cases, the C parameter is fixed to 0.002 [12] for all tested methods. We set λ = 0.9/emax, where emax is the biggest eigenvalue of Σs.