This paper describes our team's collaborative efforts in participating in the Track1 of the Diarization of Speaker and Language in Conversation Environments (DISPLACE) Challenge 2023. Our submission focuses on speaker diarization in multilingual scenarios, dealing with overlapping speech segments with significant noise ratios. To achieve our goal, we fine-tuned the parameters of two speaker diarization toolkits, Pyannote and NeMo, and retrained some components using the DISPLACE development sets and subsets from the MUSAN speech database. The experiments show promising results, we managed to make improvements over the pretrained voice activity detection (VAD) model, as well as training the Multi-scale Speaker Diarization Decoder (MSDD) by using the DISPLACE development datasets. Best systems are combined using DOVER-Lap. Our approach achieves a diarization error rate (DER) of 28.97% on Phase 1 Eval set, compared to the baseline diarization error rate of 40%.