-
Notifications
You must be signed in to change notification settings - Fork 0
/
ai4-metadata.yml
34 lines (32 loc) · 1.33 KB
/
ai4-metadata.yml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
metadata_version: 2.0.0
title: Integrated Plant Protection
summary: A Torch model to classify plants.
description: |-
The module contains 4 trained models based on CNN architecture with reduced network capacity due to the size of datasets.
The pipeline is implemented in [PyTorch](https://pytorch.org/).
For rye, the first classifier in the normal approach is `rye-3` and for beets it is `beet-52`.
In the second approach, the impact of using preprocessing before the classifier was checked.
For this purpose, masks were collected and models were trained with U-Net architecture for segmentation. `rye-1` for rye and `beet-1` for beet.
Two more classifiers, namely `rye-7` and `beet-54`, were used to preprocess the corresponding models.
It was possible to train new models without and with preprocessing.
The same scheme is then used for prediction, with the difference that preprocessing is done immediately before GPU inference.
dates:
created: '2023-12-20'
updated: '2024-08-12'
links:
source_code: https://github.com/ai4os-hub/integrated-plant-protection
docker_image: ai4oshub/integrated-plant-protection
ai4_template: ai4-template/1.9.9
tags:
- deep learning
- image classification
tasks:
- Computer Vision
- Classification
categories:
- AI4 pre trained
- AI4 inference
libraries:
- PyTorch
data-type:
- Image