Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,39 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
library_name: keras
|
| 4 |
+
tags:
|
| 5 |
+
- image-classification
|
| 6 |
+
- image-segmentation
|
| 7 |
---
|
| 8 |
+
|
| 9 |
+
## Model Description
|
| 10 |
+
### Keras Implementation of Point cloud classification with PointNet
|
| 11 |
+
|
| 12 |
+
This repo contains the trained model of [Point cloud classification with PointNet](https://keras.io/examples/vision/pointnet/).
|
| 13 |
+
|
| 14 |
+
The full credit goes to: [David Griffiths](https://dgriffiths3.github.io/)
|
| 15 |
+
|
| 16 |
+
## Intended uses & limitations
|
| 17 |
+
- As stated in the paper, PointNet is 3D perception model, applying deep learning to point clouds for object classification and scene semantic segmentation.
|
| 18 |
+
- PointNet takes raw point cloud data as input, which is typically collected from either a lidar or radar sensor.
|
| 19 |
+
|
| 20 |
+
## Training and evaluation data
|
| 21 |
+
- The dataset used for training is ModelNet10, the smaller 10 class version of the ModelNet40 dataset.
|
| 22 |
+
|
| 23 |
+
## Training procedure
|
| 24 |
+
### Training hyperparameter
|
| 25 |
+
The following hyperparameters were used during training:
|
| 26 |
+
- optimizer: 'adam'
|
| 27 |
+
- loss: 'sparse_categorical_crossentropy'
|
| 28 |
+
- epochs: 20
|
| 29 |
+
- batch_size: 32
|
| 30 |
+
- learning_rate: 0.001
|
| 31 |
+
|
| 32 |
+
## Model Plot
|
| 33 |
+
|
| 34 |
+
<details>
|
| 35 |
+
<summary>View Model Plot</summary>
|
| 36 |
+
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
</details>
|