MaGGIe introduces an efficient framework using Cross-Attention, Self-Attention, and Sparse Convolutions for mask-guided instance matting, ensuring high accuracyMaGGIe introduces an efficient framework using Cross-Attention, Self-Attention, and Sparse Convolutions for mask-guided instance matting, ensuring high accuracy

MaGGIe Architecture: Efficient Mask-Guided Instance Matting

Abstract and 1. Introduction

  1. Related Works

  2. MaGGIe

    3.1. Efficient Masked Guided Instance Matting

    3.2. Feature-Matte Temporal Consistency

  3. Instance Matting Datasets

    4.1. Image Instance Matting and 4.2. Video Instance Matting

  4. Experiments

    5.1. Pre-training on image data

    5.2. Training on video data

  5. Discussion and References

\ Supplementary Material

  1. Architecture details

  2. Image matting

    8.1. Dataset generation and preparation

    8.2. Training details

    8.3. Quantitative details

    8.4. More qualitative results on natural images

  3. Video matting

    9.1. Dataset generation

    9.2. Training details

    9.3. Quantitative details

    9.4. More qualitative results

3. MaGGIe

We introduce our efficient instance matting framework guided by instance binary masks, structured into two parts. The first Sec. 3.1 details our novel architecture to maintain accuracy and efficiency. The second Sec. 3.2 describes our approach for ensuring temporal consistency across frames in video processing.

3.1. Efficient Masked Guided Instance Matting

\

\ In cross-attention (CA), Q and (K, V) originate from different sources, whereas in self-attention (SA), they share similar information.

\

\ Figure 2. Overall pipeline of MaGGIe. This framework processes frame sequences I and instance masks M to generate per-instance alpha mattes A′ for each frame. It employs progressive refinement and sparse convolutions for accurate mattes in multi-instance scenarios, optimizing computational efficiency. The subfigures on the right illustrate the Instance Matte Decoder and the Instance Guidance, where we use mask guidance to predict coarse instance mattes and guide detail refinement by deep features, respectively. (Optimal in color and zoomed view).

\ where {; } denotes concatenation along the feature dimension, and G is a series of sparse convolutions with sigmoid activation.

\

\

:::info Authors:

(1) Chuong Huynh, University of Maryland, College Park (chuonghm@cs.umd.edu);

(2) Seoung Wug Oh, Adobe Research (seoh,jolee@adobe.com);

(3) Abhinav Shrivastava, University of Maryland, College Park (abhinav@cs.umd.edu);

(4) Joon-Young Lee, Adobe Research (jolee@adobe.com).

:::


:::info This paper is available on arxiv under CC by 4.0 Deed (Attribution 4.0 International) license.

:::

\

Market Opportunity
Mask Network Logo
Mask Network Price(MASK)
$0.5593
$0.5593$0.5593
-1.07%
USD
Mask Network (MASK) Live Price Chart
Disclaimer: The articles reposted on this site are sourced from public platforms and are provided for informational purposes only. They do not necessarily reflect the views of MEXC. All rights remain with the original authors. If you believe any content infringes on third-party rights, please contact service@support.mexc.com for removal. MEXC makes no guarantees regarding the accuracy, completeness, or timeliness of the content and is not responsible for any actions taken based on the information provided. The content does not constitute financial, legal, or other professional advice, nor should it be considered a recommendation or endorsement by MEXC.

You May Also Like

U.S. Court Finds Pastor Found Guilty in $3M Crypto Scam

U.S. Court Finds Pastor Found Guilty in $3M Crypto Scam

The post U.S. Court Finds Pastor Found Guilty in $3M Crypto Scam appeared on BitcoinEthereumNews.com. Crime 18 September 2025 | 04:05 A Colorado judge has brought closure to one of the state’s most unusual cryptocurrency scandals, declaring INDXcoin to be a fraudulent operation and ordering its founders, Denver pastor Eli Regalado and his wife Kaitlyn, to repay $3.34 million. The ruling, issued by District Court Judge Heidi L. Kutcher, came nearly two years after the couple persuaded hundreds of people to invest in their token, promising safety and abundance through a Christian-branded platform called the Kingdom Wealth Exchange. The scheme ran between June 2022 and April 2023 and drew in more than 300 participants, many of them members of local church networks. Marketing materials portrayed INDXcoin as a low-risk gateway to prosperity, yet the project unraveled almost immediately. The exchange itself collapsed within 24 hours of launch, wiping out investors’ money. Despite this failure—and despite an auditor’s damning review that gave the system a “0 out of 10” for security—the Regalados kept presenting it as a solid opportunity. Colorado regulators argued that the couple’s faith-based appeal was central to the fraud. Securities Commissioner Tung Chan said the Regalados “dressed an old scam in new technology” and used their standing within the Christian community to convince people who had little knowledge of crypto. For him, the case illustrates how modern digital assets can be exploited to replicate classic Ponzi-style tactics under a different name. Court filings revealed where much of the money ended up: luxury goods, vacations, jewelry, a Range Rover, high-end clothing, and even dental procedures. In a video that drew worldwide attention earlier this year, Eli Regalado admitted the funds had been spent, explaining that a portion went to taxes while the remainder was used for a home renovation he claimed was divinely inspired. The judgment not only confirms that INDXcoin qualifies as a…
Share
BitcoinEthereumNews2025/09/18 09:14
MSCI’s Proposal May Trigger $15B Crypto Outflows

MSCI’s Proposal May Trigger $15B Crypto Outflows

MSCI's plan to exclude crypto-treasury companies could cause $15B outflows, impacting major firms.
Share
CoinLive2025/12/19 13:17
This U.S. politician’s suspicious stock trade just returned over 200% in weeks

This U.S. politician’s suspicious stock trade just returned over 200% in weeks

The post This U.S. politician’s suspicious stock trade just returned over 200% in weeks appeared on BitcoinEthereumNews.com. United States Representative Cloe Fields has seen his stake in Opendoor Technologies (NASDAQ: OPEN) stock return over 200% in just a matter of weeks. According to congressional trade filings, the lawmaker purchased a stake in the online real estate company on July 21, 2025, investing between $1,001 and $15,000. At the time, the stock was trading around $2 and had been largely stagnant for months. Receive Signals on US Congress Members’ Stock Trades Stocks Stay up-to-date on the trading activity of US Congress members. The signal triggers based on updates from the House disclosure reports, notifying you of their latest stock transactions. Enable signal The trade has since paid off, with Opendoor surging to $10, a gain of nearly 220% in under two months. By comparison, the broader S&P 500 index rose less than 5% during the same period. OPEN one-week stock price chart. Source: Finbold Assuming he invested a minimum of $1,001, the purchase would now be worth about $3,200, while a $15,000 stake would have grown to nearly $48,000, generating profits of roughly $2,200 and $33,000, respectively. OPEN’s stock rally Notably, Opendoor’s rally has been fueled by major corporate shifts and market speculation. For instance, in August, the company named former Shopify COO Kaz Nejatian as CEO, while co-founders Keith Rabois and Eric Wu rejoined the board, moves seen as a return to the company’s early innovative spirit.  Outgoing CEO Carrie Wheeler’s resignation and sale of millions in stock reinforced the sense of a new chapter. Beyond leadership changes, Opendoor’s surge has taken on meme-stock characteristics. In this case, retail investors piled in as shares climbed, while short sellers scrambled to cover, pushing prices higher.  However, the stock is still not without challenges, where its iBuying model is untested at scale, margins are thin, and debt tied to…
Share
BitcoinEthereumNews2025/09/18 04:02