Local Minima Free Parameterized Appearance Models
Description
Parameterized Appearance Models (PAMs) (e.g. Eigen-tracking, Active Appearance Models, Morphable Models) are commonly used to model the appearance and shape variation of objects in images. While PAMs have numerous advantages relative to alternate approaches, they have at least two drawbacks. First, they are especially prone to local minima in the fitting process. Second, often few if any of the local minima of the cost function correspond to acceptable solutions. To solve these problems, this paper proposes a method to learn a cost function by explicitly optimizing that the local minima occur at and only at the places corresponding to the correct fitting parameters. To the best of our knowledge, this is the first paper to address the problem of learning a cost function to explicitly model local properties of the error surface to fit PAMs. Synthetic and real examples show improvement in alignment performance in comparison with traditional approaches.
| Slides | |
| 0:00 | Local Minima Free Parameterized Appearance Models |
| 0:46 | Image alignment |
| 1:13 | Some work in image alignment |
| 1:59 | Image alignment as an optimization problem |
| 4:42 | Issues of gradient‐based optimization |
| 7:02 | Local minima problems in image alignment |
| 8:19 | Quadratic cost function part1 |
| 8:52 | Quadratic cost function part2 |
| 9:05 | Training data |
| 9:20 | Multiple training images |
| 9:55 | 1st desired property of the cost function |
| 10:22 | 2nd desired property of the cost function |
| 11:32 | Enforcing the 2nd desired property |
| 12:53 | The learning problem part1 |
| 14:03 | The learning problem part2 |
| 14:45 | Weighted template alignment |
| 15:46 | Experiment with the Gaussian |
| 16:05 | Error surfaces |
| 16:33 | Active Appearance Models (AAMs) |
| 17:14 | Energy function in AAMs |
| 18:48 | Learning AAM energy function |
| 20:04 | Experiments with Multi‐PIE database |
| 20:32 | Results of weighted PCA basis |
| 22:07 | Summary |
| 22:48 | Thank you |
| 23:00 | - Questions |
Lecture rating
| People found this lecture: | ||
| Worth seeing | ||
| because it is: | ||
| Valuable and informative | ||
| Well presented | ||
| Easily understandable | ||
| Acceptably recorded | ||
| You need to login to cast your vote. | ||
Report a problem or upload files
If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
Related content
Link this page
Would you like to put a link to this lecture on your homepage?Go ahead! Copy the HTML snippet !





