|
|
--- |
|
|
'[object Object]': null |
|
|
license: mit |
|
|
datasets: |
|
|
- ddecosmo/lanternfly_training_dataset |
|
|
language: |
|
|
- en |
|
|
base_model: |
|
|
- google/efficientnet-b1 |
|
|
--- |
|
|
|
|
|
# Model Card for {{ model_id | default("Model ID", true) }} |
|
|
|
|
|
<!-- Provide a quick summary of what the model is/does. --> |
|
|
|
|
|
This is a fine tuned version of an EfficientNetB1 model trained on lanternfly, other insects, and general photos |
|
|
for classification. |
|
|
|
|
|
## Model Details |
|
|
|
|
|
### Model Description |
|
|
|
|
|
This model uses an EfficientNEtB1 with an Adam optimizer, mulit-class accuracy, and cross entropy loss. |
|
|
|
|
|
- **Developed by:** Devin DeCosmo |
|
|
- **Model type:** Image Classifier |
|
|
- **Language(s) (NLP):** English |
|
|
- **License:** MIT |
|
|
- **Finetuned from model:** EfficientNetB1 |
|
|
|
|
|
## Uses |
|
|
|
|
|
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. --> |
|
|
|
|
|
This model is used for classifying lanterfly photos vs other insects and non insect photos. |
|
|
|
|
|
### Direct Use |
|
|
|
|
|
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. --> |
|
|
|
|
|
The direct use is classiying photos within the 3 classes provided. Lanternfly, other insect, and non insect classes. |
|
|
|
|
|
### Out-of-Scope Use |
|
|
|
|
|
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. --> |
|
|
|
|
|
This could be expanded to additional insect classes to expand range tracking capabilities. |
|
|
|
|
|
## Bias, Risks, and Limitations |
|
|
|
|
|
<!-- This section is meant to convey both technical and sociotechnical limitations. --> |
|
|
|
|
|
This model is trained off a subset of lanternfly, insect, and non lanternfly images. |
|
|
The dataset is a moderate size with a large number of augmented values. |
|
|
It is accurate to 95% within testing and validation but there are edge cases |
|
|
not included in the dataset that cause errors. |
|
|
|
|
|
This includes insects in locations not included in training data and outdoor scenes with different lighting. |
|
|
The dataset should be expanded or the model should be changed to improve it. |
|
|
|
|
|
### Recommendations |
|
|
|
|
|
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. --> |
|
|
|
|
|
The gaps found within this dataset, other insects and other lighting conditions, mean this model cannot be trusted in |
|
|
all novel environment. Expanding this dataset or altering this model to include technique like |
|
|
blob identification would mitigate this issue. |
|
|
|
|
|
|
|
|
## Training Details |
|
|
|
|
|
### Training Data |
|
|
|
|
|
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. --> |
|
|
|
|
|
ddecosmo/lanternfly_training_dataset |
|
|
|
|
|
This is the training dataset used. |
|
|
|
|
|
### Training Procedure |
|
|
|
|
|
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. --> |
|
|
This model was trained with an AutoML process with accuracy as the main metrics. The modelw as trained over 20 epochs with a batch size of 32 images. |
|
|
|
|
|
|
|
|
#### Training Hyperparameters |
|
|
|
|
|
This model used an Adam optimizer, mulit-class accuracy, and cross entropy loss. |
|
|
|
|
|
|
|
|
## Evaluation |
|
|
|
|
|
<!-- This section describes the evaluation protocols and provides the results. --> |
|
|
|
|
|
### Testing Data, Factors & Metrics |
|
|
|
|
|
#### Testing Data |
|
|
|
|
|
<!-- This should link to a Dataset Card if possible. --> |
|
|
ddecosmo/lanternfly_training_dataset |
|
|
The testing data was the 'original' split, the original and 3rd party images in this set. |
|
|
|
|
|
#### Factors |
|
|
|
|
|
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. --> |
|
|
|
|
|
This dataset is evaluating whether the food is Lanternfly, "0", or Other Insect, "1", or Non Insect "2". |
|
|
|
|
|
#### Metrics |
|
|
|
|
|
<!-- These are the evaluation metrics being used, ideally with a description of why. --> |
|
|
|
|
|
The testing metric used was accuracy to ensure the highest accuracy of the model possible. |
|
|
|
|
|
|
|
|
### Results |
|
|
|
|
|
After training with the initial dataset, this model reached an accuracy of 95% in validation. |
|
|
|
|
|
#### Summary |
|
|
|
|
|
This model reached a high accuracy with our current model. |
|
|
The large size of the dataset allowed for a large amount of training. |
|
|
After training, it was found the training dataset had gaps, causing edge case failures |
|
|
that fell outside the bounds of the original dataset. |
