1
$\begingroup$

I am working on a project in which I will be performing object detection on deformed objects. Unfortunately, there isn't enough data sets to train them on some neural network. I am looking for reference on computer vision algorithms that does not require learning or training on datasets. Rather, topics arising from "structure from motion" that are popular within the computer vision community.

Some note: The object detection will take place in an open environment rather close environment.

$\endgroup$

1 Answer 1

0
+100
$\begingroup$

"... I am looking for reference on computer vision algorithms that does not require learning or training on datasets. Rather, topics arising from "structure from motion" ...".

Here is a list of papers, blogs, and code that you might find helpful:

  • The website PapersWithCode's search engine returns several hits for "NRSfM", a few of the projects specifically mention limited datasets:

    1. "Deep Interpretable Non-Rigid Structure from Motion", (Feb 28 2019), by Chen Kong and Simon Lucey. GitHub.

      "Once the network's weights are estimated (for a non-rigid object) we show how our approach can effectively recover 3D shape from a single image -- outperforming comparable methods that rely on direct 3D supervision.".

    2. "Local Non-Rigid Structure-From-Motion From Diffeomorphic Mappings", (CVPR 2020), by Shaifali Parashar, Mathieu Salzmann, and Pascal Fua. GitHub.

      "We show that it is nevertheless sufficient to establish local correspondences between the surface in two different images and therefore to perform point-wise reconstruction using only first-order derivatives. ".

    3. "MHR-Net: Multiple-Hypothesis Reconstruction of Non-Rigid Shapes from 2D Views", (July 19, 2022), by Haitian Zeng, Xin Yu, Jiaxu Miao, and Yi Yang. (No code implementations yet).

      "We propose MHR-Net, a novel method for recovering Non-Rigid Shapes from Motion (NRSfM). MHR-Net aims to find a set of reasonable reconstructions for a 2D view, and it also selects the most likely reconstruction from the set. To deal with the challenging unsupervised generation of non-rigid shapes, we develop a new Deterministic Basis and Stochastic Deformation scheme in MHR-Net.".

  • arXiv's search engine returns 31 hits for "NRSfM", a few of the papers are:

Some other papers are:

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .