Conditional image generation has gained significant attention for its ability to personalize content. However, the field faces challenges in developing task-agnostic, reliable, and explainable evaluation metrics. This paper introduces CIGEval, a unified agentic framework for comprehensive evaluation of conditional image generation tasks. CIGEval utilizes large multimodal models (LMMs) as its core, integrating a multi-functional toolbox and establishing a fine-grained evaluation framework. Additionally, we synthesize evaluation trajectories for fine-tuning, empowering smaller LMMs to autonomously select appropriate tools and conduct nuanced analyses based on tool outputs. Experiments across seven prominent conditional image generation tasks demonstrate that CIGEval (GPT-4o version) achieves a high correlation of 0.4625 with human assessments, closely matching the inter-annotator correlation of 0.47. Notably, when implemented with 7B open-source LMMs using only 2.3K training trajectories, CIGEval surpasses the previous GPT-4o-based state-of-the-art method. These findings indicate that CIGEval holds great potential for automating evaluation of image generation tasks while maintaining human-level reliability.

The content-centric data sharing in Intelligent Transportation Systems (ITS) becomes increasingly imperative to improve road safety and traffic efficiency. In order to protect the privacy of vehicle users, data is generally encrypted prior to being shared, whereas it hinders data searching and utilization. Proxy re-encryption with equivalence test (PREET) has been proposed to test the equivalence between encrypted messages, and to delegate the decryption right of searched data to the specified user. However, existing PREET schemes primarily focus on the equivalence test between two ciphertexts, which aren't suitable for the most practical scenarios that more than two ciphertexts need to be verified, due to the exposure of users' extra information and the redundant computation. In this paper, to tackle these problems, we present a lightweight certificateless proxy re-encryption with multi-ciphertext equivalence test (CL-PREMET) for shared data to address the data search and authorized utilization over encrypted data in ITS. In our design, the users are allowed to efficiently and accurately search over massive encrypted data based on flexible multi-ciphertext analysis and to utilize their own keys rather than data owner's key to decrypt the searched data after being authorized, without revealing any private and sensitive information. The CL-PREMET achieves the IND-CCA security and the OW-CCA security. Furthermore, the simulation experiment displays that the CL-PREMET is feasible. Compared with existing PREET schemes, in addition to being suitable for multi-ciphertext equivalence analysis, our scheme leverages the certificateless public key cryptography (CL-PKC) to avoid certificate management and key escrow problems.

The low throughput of traditional blockchain can hardly support the demand of high-frequency and high-concurrent application scenarios, which severely restricts the practical application of blockchain and becomes a bottleneck for blockchain development. Sharding is a potential solution to this issue. Existing sharding schemes mostly use complicated PBFT or its variants consensus protocols, with diminishing consensus performance as the number of nodes grows, and global consistency of transaction data is a challenge due to parallel block generation. Aiming at the above problems, we propose an efficient and scalable sharding blockchain system named FortunChain that comprises a verification chain and multiple sharding chains. In the FortunChain, a novel Tyche consensus protocol is designed to select the set of sharding verification nodes and accounting nodes probabilistically leveraging the newly constructed elliptic curve-based verifiable random function (EC-VRF), which ensures the efficiency, unpredictability and security of node selection. Multi-period, fast and dynamic sharding accounting is implemented in the sharding chain to improve the performance of transaction processing. After that, the verification chain is responsible for the secondary verification by applying the batch verification algorithm to the blocks submitted by the shards and then generating the verification block. Each shard stores the sharding state by synchronizing the verification chain to realize the state sharding and ensure the global consistency of the transaction data. Simulation results show that the throughput of FortunChain can reach approximately 6540tx/s and the estimated latency is 8.59s, which demonstrates its efficiency and feasibility.