Trees Counting | ||||||||
Palm trees counting over different countries (Ghana, Indonesia, Papua, Sri Lanka, and Brazil) (Source: https://feds.ae/saving-time-and-effort-training-an-ai-to-count-palm-trees/) | ||||||||
Product Category | ||||||||
|
| |||||||
Financial Domains | ||||||||
| ||||||||
User requirements | ||||||||
UN31: Need to link tree planting parcels to estimate the number of trees planted | ||||||||
Description | ||||||||
Tree counting require VHR satellite imagery to accurately and efficiently determine the number of trees within a specified area. This approach offers significant advantages over traditional ground-based methods, as it enables rapid, cost-effective, and large-scale tree counting without the need for time-consuming field surveys. This data, coupled with the ability to monitor changes over time, aids in assessing the value of forestry assets, estimating timber volume for investment or insurance purposes, and evaluating the environmental impact of forestry investments. | ||||||||
Spatial Coverage Target | ||||||||
Asset level | ||||||||
Data Throughput | ||||||||
Rapid tasking Data availability |
|
| ||||||
PRODUCT SPECIFICATIONS | ||||||||
Main processing steps | The process starts by inspecting optical VHR imagery (≤ 0.5 m) to identify a sample of individual trees to build a training dataset for a deep learning model. Then a deep learning model to be trained in optimoum way to detect the individual tree and, subsequently, identify the number of trees. | |||||||
Input data sources | Optical: VHR based on the availability like Pleiades 1A/1B & NEO, WorldView2&3, and SPOT6/7 Radar: N.A Supporting data: N.A | |||||||
Accessibility | Optical VHR imagery: commercially available on demand from EO service providers. | |||||||
Spatial resolution | Optical VHR: ≤ 0.5 m | |||||||
Frequency (Temporal resolution) | Optical VHR: Sub-daily to Daily | |||||||
Latency | < 1 Day | |||||||
Geographical scale coverage | Globally | |||||||
Delivery/ output format | Data type: Raster File format: GeoTIFF | |||||||
Accuracies | Thematic accuracy: 70-80% Spatial accuracy: 1.5-2 pixels of input data | |||||||
Constraints and limitations |
| |||||||
User's level of knowledge and skills to extract information and perform further analysis on the EO products. | Skills: Essential Knowledge: Essential |
P08: Trees counting | |
Maturity score | |
Mean: 2.4 | STD: 0.70 |
Constraints and limitations · Cloud presence · Machine learning model uncertainty | |
Relevant user needs UN31: Need to link tree planting parcels to estimate the number of trees planted. | |
R&D gaps · Cost of Very High Resolution (VHR) satellite imagery which is essential for the product. · Global inconsistency due to the diversity of tree species. · Limitations in homogeneous forests where the trees are connected to each other. · The lack of local in-situ data to train and validate the models. · Lack of spectral resolution to differentiate between tree species | |
Potential improvements drivers · Advances in AI models to detect and count individual trees. · Datasets for training and validating the models. · Price models for commercial EO data. · Fusion of hyperspectral and multispectral EO data. | |
Utilisation level review | |
Utilisation score | |
Mean: 2.14 | STD: 0.64 |
No utilisation: Unawareness of the existence of this EO product Low utilisation · Higher cost of using the commercial EO product. · The current method (manually counting for a sample area and multiplying up to estimate the whole area) is considered good enough in terms of accuracy, reliability, and price. · Ground truth data is not sufficient for counting individual trees. Medium utilisation · Unawareness of the existence of the best available commercial EO product with better specifications. · Higher cost of using the best available commercial EO product . High utilisation | |
Critical gaps related to relevant user needs | |
Utilisation gap UN31: Need to link tree planting parcels to estimate the number of trees planted |
This page has no comments.