Sensia

HOME　　Image Processing System　　Detecting Missing Ingredients. Bento box Ingredient Inspection Technology.

Detecting Missing Ingredients. Bento box Ingredient Inspection Technology.

Judgment screen

In manual bento box preparation, issues such as missing items or incorrect placement can occur. This technology captures images of bento boxes and uses AI (object detection) and image processing to identify the contents, detecting any missing items or placement errors within the bento box compartments. The AI can also detect variations in the individual food items. The registration process of bento box configurations into the AI is simplified, allowing for compatibility with various containers and ingredients.

What Can Be Solved

  • I want to detect mistakes in manual food placement.
  • We manually check for missing items or incorrect placement, but there are instances where mistakes are overlooked.

Examples of Detection

  • Image of pickles being out of stock

    Detection of Missing Pickles

  • Image of cherry tomatoes placed in the wrong place

    Incorrect Placement of Cherry Tomatoes

  • Image of a shortage of noodle soup

    Detection of Missing Noodle Sauce

Features

Detection of side dishes using AI (object detection)

Image of detecting side dishes

AI detects the type and position of side dishes. It performs relative detection with respect to the container and determines whether the correct side dishes are placed in the designated positions, thus detecting missing items or placement errors. AI-based detection also accounts for shape variations due to individual differences.

Freely configurable detection range

Judgment range setting example

Users can freely set the detection range. It can accommodate various container shapes and side dish arrangements. Set the range relative to the container and configure the corresponding side dishes.

Simple AI training tasks

Learning image

The types and positions of side dishes are learned based on the set detection range. After configuring the detection range, learning can be done by capturing images of correctly assembled items.

Detection of items other than side dishes is also possible.

Image of food to be judged other than side dishes

By training the AI, it is also possible to detect accessories such as soy sauce containers and chopsticks. This can be used to verify correct packaging of the bento.

Automatic detection

Automatic judgment image

Automatic detection allows for further simplification of the workflow. By detecting the container, the system automatically starts the assessment and displays the results.

AI Training Procedure

To perform detection using AI (object detection), it is necessary to create annotated data that indicates the types and positions of the ingredients on the images and to train the AI based on this annotated data. This technology simplifies the process by pre-setting the types and positions of the ingredients on the container, allowing for automatic annotation of captured images.

Learning Procedure
  1. Take a photograph of the lunchbox container.
  2. Set the types and areas of the side dishes for the lunchbox container.
  3. Take multiple photos of the normal lunchbox. Based on the settings from step ②, the system will automatically annotate which side dishes are located where in the lunchbox container.
  4. By using the data from step ③ for AI training, the system will be able to detect and evaluate the side dishes.