| Download | - View author's version: Transferring knowledge for food image segmentation using transformers and convolutions (PDF, 2.5 MiB)
|
|---|
| DOI | Resolve DOI: https://doi.org/10.48550/arXiv.2306.09203 |
|---|
| Author | Search for: Sinha, Grant; Search for: Parmar, Krish; Search for: Azimi, Hilda1; Search for: Tai, Amy; Search for: Chen, Yuhao; Search for: Wong, Alexander; Search for: Xi, Pengcheng1ORCID identifier: https://orcid.org/0000-0003-3236-5234 |
|---|
| Affiliation | - National Research Council of Canada. Digital Technologies
|
|---|
| Format | Text, Preprint |
|---|
| Conference | The 2nd Workshop on Computer Vision in the Wild, June 19, 2023, Vancouver, British Columbia, Canada |
|---|
| Abstract | Food image segmentation is an important task that has ubiquitous applications, such as estimating the nutritional value of a plate of food. Although machine learning models have been used for segmentation in this domain, food images pose several challenges. One challenge is that food items can overlap and mix, making them difficult to distinguish. Another challenge is the degree of inter-class similarity and intra-class variability, which is caused by the varying preparation methods and dishes a food item may be served in. Additionally, class imbalance is an inevitable issue in food datasets. To address these issues, two models are trained and compared, one based on convolutional neural networks and the other on Bidirectional Encoder representation for Image Transformers (BEiT). The models are trained and valuated using the FoodSeg103 dataset, which is identified as a robust benchmark for food image segmentation. The BEiT model outperforms the previous state-of the-art model by achieving a mean intersection over union of 49.4 on FoodSeg103. This study provides insights into transfering knowledge using convolution and Transformer based approaches in the food image domain. |
|---|
| Publication date | 2023-06-19 |
|---|
| Publisher | Cornell University |
|---|
| Licence | |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | No |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | a44c3198-ab99-4640-885c-f881841f369c |
|---|
| Record created | 2023-11-03 |
|---|
| Record modified | 2023-11-08 |
|---|
