Age | Commit message (Collapse) | Author | |
---|---|---|---|
2021-03-15 | Remove redundant wrapper given by dataloader | Jordan Gong | |
2021-03-15 | Fix redundant gallery_dataset_meta assignment | Jordan Gong | |
2021-03-15 | Support transforming on training datasets | Jordan Gong | |
2021-03-14 | Bug fix when transforming and new config | Jordan Gong | |
2021-03-14 | Fix unbalanced datasets | Jordan Gong | |
2021-03-12 | Fix a typo when record none-zero counts | Jordan Gong | |
2021-03-12 | Make evaluate method static | Jordan Gong | |
2021-03-12 | Code refactoring | Jordan Gong | |
1. Separate FCs and triplet losses for HPM and PartNet 2. Remove FC-equivalent 1x1 conv layers in HPM 3. Support adjustable learning rate schedulers | |||
2021-03-10 | Bug fixes | Jordan Gong | |
1. Resolve reference problems when parsing dataset selectors 2. Transform gallery using different models | |||
2021-03-08 | Bump up package version | Jordan Gong | |
2021-03-05 | Bump up torch version | Jordan Gong | |
2021-03-04 | Replace detach with no_grad in evaluation | Jordan Gong | |
2021-03-04 | Set seed for reproducibility | Jordan Gong | |
2021-03-03 | Add L2 penalty to global | Jordan Gong | |
2021-03-02 | Record learning rate every step | Jordan Gong | |
2021-03-02 | Fix bugs in new scheduler | Jordan Gong | |
2021-03-01 | New scheduler and new config | Jordan Gong | |
2021-03-01 | Change flat distance calculation method | Jordan Gong | |
2021-03-01 | Move pairs variable to local | Jordan Gong | |
2021-03-01 | Remove identical sample in Batch All case | Jordan Gong | |
2021-02-28 | Implement sum of loss default in [1] | Jordan Gong | |
[1]A. Hermans, L. Beyer, and B. Leibe, “In defense of the triplet loss for person re-identification,” arXiv preprint arXiv:1703.07737, 2017. | |||
2021-02-28 | Log n-ile embedding distance and norm | Jordan Gong | |
2021-02-28 | Modify default parameters | Jordan Gong | |
1. Change ReLU to Leaky ReLU in decoder 2. Add 8-scale-pyramid in HPM | |||
2021-02-28 | Bump up version for tqdm | Jordan Gong | |
2021-02-27 | Implement Batch Hard triplet loss and soft margin | Jordan Gong | |
2021-02-26 | Update default config | Jordan Gong | |
2021-02-26 | Fix predict function | Jordan Gong | |
2021-02-21 | Remove FConv blocks | Jordan Gong | |
2021-02-20 | Separate triplet loss from model | Jordan Gong | |
2021-02-19 | Allow evaluate unfinished model | Jordan Gong | |
2021-02-19 | Bump up tqdm version | Jordan Gong | |
2021-02-18 | Implement adjustable input size and change some default configs | Jordan Gong | |
2021-02-18 | Remove 1x1 conv layers when not used | Jordan Gong | |
2021-02-18 | Decode mean appearance feature | Jordan Gong | |
2021-02-18 | Decode mean appearance feature | Jordan Gong | |
2021-02-17 | Fix type hints and add constrains to height and width | Jordan Gong | |
2021-02-17 | Add new preprocess script | Jordan Gong | |
2021-02-16 | Split transform and evaluate method | Jordan Gong | |
2021-02-15 | Revert "Memory usage improvement" | Jordan Gong | |
This reverts commit be508061 | |||
2021-02-14 | Memory usage improvement | Jordan Gong | |
This update separates input data to two batches, which reduces ~30% memory usage. | |||
2021-02-14 | Prepare for DataParallel | Jordan Gong | |
2021-02-13 | Sum gallery dimension instead of all dimensions | Jordan Gong | |
2021-02-10 | Implement new sampling technique mentioned in GaitPart[1] | Jordan Gong | |
[1]C. Fan et al., “GaitPart: Temporal Part-Based Model for Gait Recognition,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 14225–14233. | |||
2021-02-10 | Save scheduler state_dict | Jordan Gong | |
2021-02-09 | Improve performance when disentangling | Jordan Gong | |
This is a HUGE performance optimization, up to 2x faster than before. Mainly because of the replacement of randomized for-loop with randomized tensor. | |||
2021-02-09 | Some optimizations | Jordan Gong | |
1. Scheduler will decay the learning rate of auto-encoder only 2. Write learning rate history to tensorboard 3. Reduce image log frequency | |||
2021-02-08 | Code refactoring, modifications and new features | Jordan Gong | |
1. Decode features outside of auto-encoder 2. Turn off HPM 1x1 conv by default 3. Change canonical feature map size from `feature_channels * 8 x 4 x 2` to `feature_channels * 2 x 16 x 8` 4. Use mean of canonical embeddings instead of mean of static features 5. Calculate static and dynamic loss separately 6. Calculate mean of parts in triplet loss instead of sum of parts 7. Add switch to log disentangled images 8. Change default configuration | |||
2021-01-23 | Remove the third term in canonical consistency loss | Jordan Gong | |
2021-01-23 | Add late start support for non-disentangling parts | Jordan Gong | |
2021-01-23 | Transform all frames together in evaluation | Jordan Gong | |