← All projects · Computer Vision · AI Engineering
Generation & Restoration
Muhammad Hammas · AI Engineer · Master's in Artificial Intelligence
Image-to-image restoration and translation, plus generation from text. Every project ships the images it produced alongside the number it scored. 15 projects.
Cloud removal
Fine-tuned from the published RICE2 generator for 60 epochs on 1,600 training pairs
23.6964 dB PSNR (test)Show the resultRemove haze and cloud from aerial imagery.
- Fine-tuned from the published RICE2 generator for 60 epochs on 1,600 training pairs.
- Train 24.13 / val 23.97 / test 23.70 - a tight spread across all three splits.
split PSNR train n=1600 24.1264 val n=300 23.9666 test n=100 23.6964 
Colourisation
Trained from scratch for 100 epochs on 2,280 images, predicting the ab channels from L in Lab space
21.9107 dB PSNR (test)Show the resultPredict colour (ab) from luminance (L) in Lab space.
- Trained from scratch for 100 epochs on 2,280 images, predicting the ab channels from L in Lab space.
- The dataset was filtered for colour content: near-grey and low-saturation images were dropped when it was built.
split PSNR train n=428 22.5389 val n=428 22.2671 test n=142 21.9107 
CUT: horse to zebra
100 epochs on the horse2zebra set
Show the result- 100 epochs on the horse2zebra set. CUT is unpaired, so the result is shown as images rather than a pixel metric.
- Budget matched to cyclegan-h2z, which trains on the same data.
CycleGAN: apple to orange
100 epochs on the apple2orange set, 50 at the base rate then 50 linearly decayed - the published schedule
-0.0958 negative cycle-consistency L1 on held-out images (val)Show the result- 100 epochs on the apple2orange set, 50 at the base rate then 50 linearly decayed - the published schedule.
- Samples show the same inputs translated at epochs 5 through 100.
split PSNR val n=60 -0.0958 
CycleGAN: summer to winter
30 epochs on the summer2winter_yosemite set, not the 100 the other CycleGAN runs used
-0.1049 negative cycle-consistency L1 on held-out images (val)Show the result- 30 epochs on the summer2winter_yosemite set, not the 100 the other CycleGAN runs used. The run was stopped early: the GPU dropped to half its clock (1035 of 2100 MHz, 46 of 110 W) and an epoch went from 7 to 29 minutes, which put the remaining 70 epochs at about 34 hours for a cycle-consistency gain of 0.009 per 10 epochs.
- Samples show the same inputs translated at epochs 5 through 30, so the trend over training is visible even though the budget is shorter than apple2orange's.
split PSNR val n=60 -0.1049 
Document shadow removal
Fine-tuned for 56 epochs on 800 training pairs; the best val score came at epoch 50 and five further epochs did not beat it, so the run was stopped there
30.5951 dB PSNR (test)Show the resultRemove shadows cast across a photographed document page.
- Fine-tuned for 56 epochs on 800 training pairs; the best val score came at epoch 50 and five further epochs did not beat it, so the run was stopped there.
- Train 30.67 / val 30.49 / test 30.60 - a tight spread across all three splits.
- Samples show the shadowed input, the published weights' output, this run's output and the ground truth.
split PSNR train n=800 30.6673 val n=150 30.4923 test n=50 30.5951 
ESRGAN super-resolution
Trained with L1 + adversarial
28.6938 dB PSNR (test)Show the result- Trained with L1 + adversarial.
- Fine-tuned from the published x4 checkpoint into an x6 architecture, with the upsampling stage rebuilt for a factor of 6.
- 20,000 steps on 720 training frames, evaluated on held-out val and test splits.
split PSNR train n=720 28.3642 val n=135 27.8051 test n=45 28.6938 
Face restoration
Generator fine-tuned with L1 on synthetically degraded FFHQ crops, 15,000 steps at batch 2
30.0195 dB PSNR (test)Show the resultRestore degraded face crops. Degradation is synthesised at load time.
- Generator fine-tuned with L1 on synthetically degraded FFHQ crops, 15,000 steps at batch 2.
- Sample images show the degraded input, the pretrained output and the fine-tuned output against ground truth.
split PSNR train n=150 30.0851 val n=150 29.9772 test n=50 30.0195 
MSRGAN super-resolution
Trained with L1 + adversarial
28.47 dB PSNR (test)Show the result- Trained with L1 + adversarial.
- Fine-tuned from the published x4 checkpoint into an x6 architecture, with the upsampling stage rebuilt for a factor of 6.
- 20,000 steps on 720 training frames, evaluated on held-out val and test splits.
split PSNR train n=720 28.1670 val n=135 27.5742 test n=45 28.4700 
pix2pix: facades
200 epochs on 400 training pairs - the published schedule of 100 constant plus 100 linear decay
Show the result- 200 epochs on 400 training pairs - the published schedule of 100 constant plus 100 linear decay.
- Samples show the semantic label map, the generated photograph and the ground truth at epochs 5 through 200. - pix2pix-facades-l1 is the matched control: identical architecture, data, batch, rate and schedule, with the adversarial term switched off. It scores 13.613 PSNR / 0.4369 SSIM against this run's 12.697 / 0.3928.
metric value val PSNR 12.697 
pix2pix: facades with L1
Same architecture, same data and the same 200-epoch budget as pix2pix-facades, with the adversarial term switched off
13.6132 dB PSNR (val)Show the result- Same architecture, same data and the same 200-epoch budget as pix2pix-facades, with the adversarial term switched off. The budget is held identical on purpose so the only difference between the two runs is the loss.
- Measured on the same 60-image validation split: the L1-only arm scores higher on BOTH reference metrics, +0.916 dB PSNR and +0.044 SSIM. The common framing - that L1 wins PSNR while the adversarial arm wins on perceptual quality - is only half supported here, because SSIM favours L1 too.
- So the metrics do not settle which output looks better; that is what the sample panels are for. Neither PSNR nor SSIM is a perceptual score, and no perceptual metric was run.
split PSNR val n=60 13.6132 
pix2pix: maps
200 epochs on 1,096 training pairs, same schedule as facades
Show the result- 200 epochs on 1,096 training pairs, same schedule as facades.
- Samples show the map tile, the generated aerial photograph and the ground truth across training.
metric value val PSNR 14.934 
Real-ESRGAN super-resolution
Trained with L1 + adversarial
28.257 dB PSNR (test)Show the result- Trained with L1 + adversarial.
- Fine-tuned from the published x4 checkpoint into an x6 architecture, with the upsampling stage rebuilt for a factor of 6.
- 20,000 steps on 720 training frames, evaluated on held-out val and test splits.
split PSNR train n=720 27.9199 val n=135 27.3519 test n=45 28.2570 
Real-ESRNet super-resolution
Trained with L1 only
28.3149 dB PSNR (test)Show the result- Trained with L1 only.
- Fine-tuned from the published x4 checkpoint into an x6 architecture, with the upsampling stage rebuilt for a factor of 6.
- 20,000 steps on 720 training frames, evaluated on held-out val and test splits.
split PSNR train n=720 27.9801 val n=135 27.4322 test n=45 28.3149 
Text to image
12,600 steps at batch 4, lr 1e-5, on 833 pairs
0.0513 val loss, lower is betterShow the resultGenerate an image from a text prompt.
- 12,600 steps at batch 4, lr 1e-5, on 833 pairs. The run was stopped when two successive checks gained less than 0.2% - the remaining 2,400 steps of the 15,000 budget were not spent.
- The best validation loss, 0.0513, was recorded at step 100; the loss then moved between 0.052 and 0.061 for the rest of the run. Diffusion training loss tracks sample quality loosely, so the samples are the thing to look at: samples/ holds the base model and the fine-tuned model on the same four prompts.
- No held-out test split - this set is small enough that train and validation are the only two splits used.
model tiny-SD (segmind/tiny-sd), UNet fine-tuned; VAE and text encoder frozen data Pokemon captions, 833 caption/image pairs at 512x512 
