This app can use any number of image classifiers to detect any number of target labels, then sort those images into 1 of 2 folders if the target label is detected.
To edit, add, or remove models, be sure to use the appropriate alwaysAI CLI commands to do so, in addition to adding a classifier record into the alwaysai.app.json file (see the Configuration section below).
- Sign up for alwaysAI
- Install alwaysAI tooling
This app makes use of key-values in the alwaysai.app.json file for set up and variable assignments:
| Key | Value Type | Description |
|---|---|---|
| scripts | dictionary | Required by alwaysAI's CLI to specify which file to run within a docker container |
| models | dictionary | Required by alwaysAI's CLI to determine which CV models to download and make available to the app's docker container |
| classifiers_needed_to_agree | float | Float representation of the minimum percent (100% = 1.0) of classifiers needed to make a determination if a target object was detected. So if 10 classifiers are used, a value of 0.2 here, would require at least 2 of the 10 classifiers to be in agreement before considering a target object found. Default is 0.5 |
| detected_output_folder | string | Location from the app root folder to put images with target objects into. Default is output_images/found |
| empty_output_folder | string | Location from the app root folder to put images where no target objects were detected. Default is output_images/not_found |
| classifiers | array | An array of dictionaries with classifier information. See below for expected keys for each dictionary object |
Classifier Information:
| Key | Value Type | Description |
|---|---|---|
| model_id | string | The model id as found from alwaysAI's Model Catalog. Note that there should be a duplicate of this string value in the higher level 'models' key-value pair which is used by alwaysAI's CLI tool. To add or remove models for download, see the docs here |
| confidence_level_threshold | float | Each model has a confidence_level value that's provided with each classification prediction. Many models will use a float value from 0.0-1.0 with 1.0 = 100%, but some models may have a different value based system, like squeezenet_v1_1 |
| target_labels | array of strings | List of all labels this model should be filtering / looking for. This will usually be a subset of all the available labels this model is capable of detecting. If no labels are specified nothing will be returned, in other words the detected_output_folder will be empty after processing |
Use the alwaysAI CLI to build and start this app:
Build: aai app install
Run: aai app start
NOTE: You can run this app remotely onto another device or build & run locally. Use aai app configure to select.
Docs: https://dashboard.alwaysai.co/docs/getting_started/introduction.html
Discord: https://discord.gg/alwaysai
Email: contact@alwaysai.co