# depth_estimation_MCCNN

**Repository Path**: han20121231/depth_estimation_MCCNN

## Basic Information

- **Project Name**: depth_estimation_MCCNN
- **Description**: Depth estimation on mtk data with a CNN model trained for cost computation (MCCNN)
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-07-08
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# depth_estimation_MCCNN
Depth estimation on mtk data with a CNN model trained for cost computation (MCCNN)

## Workflow
* cost computation using MCCNN
* cost aggregation by opencv blur filter
* cost optimization simply adopting winner take all
* refinement with opencv weighted median filter 

## Model
![alt text](https://github.com/leduoyang/depth_estimation_MCCNN/blob/master/Result/model.png)

## Result
![alt text](https://github.com/leduoyang/depth_estimation_MCCNN/blob/master/Result/result1.png)
![alt text](https://github.com/leduoyang/depth_estimation_MCCNN/blob/master/Result/result2.png)

source:
https://arxiv.org/pdf/1409.4326.pdf
https://github.com/jzbontar/mc-cnn